Re: sleeping process consumes CPU
On Thu, Sep 29, 2005 at 03:28:26PM +0300, Boris Zingerman wrote: Yedidyah Bar-David wrote: On Thu, Sep 29, 2005 at 11:37:34AM +0300, Boris Zingerman wrote: Hi list I'm runnig XP SP2 under vmware-5 on my linux box ( RH9 kernel 2.4.28 ) and I see that vmware-wx constantly uses 15-20% of CPU even when no application is open in XP. But what is more interesting is that top always reports this process (vmware-wx) as sleeping. How sleeping process can constantly waste CPU? If it is just an artifact of sampling that is done by top then statistically I should sometimes see it in running state. Did you try to strace it? Did you try 'top -i -d 0'? Well I straced it and it seems that it performs polling (with timeout 10msec ) on a bunch of Then I guess that's the reason. Can you try to time(1) it? E.g. time vmware boot your os etc, wait idle for a few minutes, shutdown then see how much cpu time was spend in user and in kernel. I guess both the polling itself (in user mode) and especially maybe the vm devices might use time for some reason. Maybe someone who knows can answer this: If a process does a system call (e.g. read) and the relevant driver needs a lot of cpu time, what state will the process be in? I can't think of an easy way to check this because most drivers try to use as little cpu time as possible. And I am too lazy to try and write a driver just to test this. Did you check their documentation? Do they claim you should see the machine idle if the virtual machine is idle? One more thing: Try the above time with both the BIOS's setup program waiting idle for a few minutes and with linux in single user mode. If you see significant differences it means Windows uses a lot of cpu time doing nothing. Did task manager inside it show the virtual machine is idle? -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Linux, Active Directory and TIMEZONES
On Wed, Sep 28, 2005 at 07:27:13AM +0200, Greg Pendler wrote: Hi, I've read previous posts on this issue, and tested the instructions, unfortunately nothing helps. I have Linux server that connects to AD. Actually it refuses to connect, complaining about timescew. I look at the clock on Linux and it shows 08:45 and then i look at the clock on domain controller and it also shows 08:45. I ask myself where is the timescew - servers are synchronized. Then i remember that SAMBA 3 includes *net time* command - similar to windows 200X. I run the command against the domain controller and it returns 09:45 which means that there's timezone related difference of 1 hour. Did you check the timezone on the AD server? -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: How can i tell if another process is writing to a file?
On Wed, Sep 28, 2005 at 08:05:52PM +0300, guy keren wrote: On Wed, 28 Sep 2005, Yedidyah Bar-David wrote: On Wed, Sep 28, 2005 at 06:34:09PM +0300, guy keren wrote: On Wed, 28 Sep 2005, Dvir Volk wrote: here's the situation: i have a file, and i want to open it only when no other process is writing to it. but i don't have control over the possibly writing process, so i can't do advisory or mandatory locking. basically, i want to treat a file received by scp only after the download is complete. for this specific case, you can use 'fuser', to see if there are processes holding this file open or not. you don't realy need locking for this. of course i can write the file to /tmp or somewhere and then mv it to its destination directory when it's complete. but i was wondering if there's a more elegant soultion. actually, this _is_ the more elegant solution ;) I waited for one of the experts to say this, but none did. So here I go. I never tried it myself, but there is dnotify and a replacement called inotify is in preperation. This means dnotify isn't good enough, but it works. as i understand, dnotify noties that a directory was changed. it does not notify you when a directory is NOT changed (which is what happens after the file is no longer being updated). thus, it does not sound suiteable for this task. As I said, I never tried this. But I think closing a file _is_ a change. Did you actually try this? -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: dnotify (was: Re: How can i tell if another process is writing to a file?)
On Thu, Sep 29, 2005 at 04:37:32AM +0300, guy keren wrote: On Wed, 28 Sep 2005, Yedidyah Bar-David wrote: On Wed, Sep 28, 2005 at 08:05:52PM +0300, guy keren wrote: as i understand, dnotify noties that a directory was changed. it does not notify you when a directory is NOT changed (which is what happens after the file is no longer being updated). thus, it does not sound suiteable for this task. As Isaid, I never tried this. But I think closing a file _is_ a change. Did you actually try this? just to make you happy, i went into the source tree, under Documentation/dnotify.txt. the text sais you can learn about various types of changes, that are NOT related to file opening/closing, but rather to changes done to the file or its inode. Well, I am happy, thanks. As I already said, dnotify was indeed quite limited and that's why inotify was created. I just checked and it was merged at 2.6.13, so hopefully it will soon be integrated into distributions. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Actcom without a dailer costs more
On Mon, Sep 26, 2005 at 12:25:36PM +0300, [EMAIL PROTECTED] wrote: On Sep 26, 10:26, Yedidyah Bar-David wrote: } Subject: Re: Actcom without a dailer costs more 3. Even though routers are tend to connect 24x7, yet with dynamic IPs there is still a need for significantly less IPs. Care to explain why? Not on-average, but for those who do have 24x7 connections. I do realize that if you give a static IP to every customer, there is still a significant percentage that are far from 24x7, and they would need much fewer addresses if they are dynamic. I'd also like to thank you for your sincere and interesting answers. I hope I was of some help here, even though I couldn't tell the the real price for the ISP to give a user a static IP. I was misunderstood - I did not intend to thank you for an answer to my question (and I also did not expect to get something like 5 NIS/year. Happy?), but for other things you already said and were interesting to read. Thanks, -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: [OFF TOPIC] Wiring up home network
On Fri, Sep 23, 2005 at 02:46:07PM +0300, Alexander Indenbaum wrote: Hi, I'm just finishing major shiputz at my home and now it is time to run home network wiring. From topology point of view, one of the rooms is designated as server/communication room. In the communication room there will be cable connection to the Internet with Unix router/nat server. All other bedrooms will have Ethernet sockets and wires will run to the patch panel at the communication room. [PC1] | | [PC2] |-- [patch-panel] -- [switch] -- [UNIX router] -- (INTERNET) | [PC3] | At this stage I have got tubes in walls connecting all relevant points. Here are my questions: 1. Where do I buy communication gear: a. a lot of unpressed raw Ethernet cable b. pressing equipment b. sockets and patch panel for the communication room In Tel-Aviv area, I can recommend argencom. c. switch ( this is easy to come by, but vendor/model recommendation will be appreciated ) I usually buy Edimax, as others said. Cheap, 3 years warranty, good enough. 2. How should I press the cables between sockets in the bedrooms and the patch panel in the communication room. Straight or cross? Could you describe exact order of the wires? Of course google is your friend. If you also like to keep bookmarks, google for the hardware book. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Actcom without a dailer costs more
On Wed, Sep 21, 2005 at 03:59:57AM -0700, Alex Shnitman wrote: Same story here. I signed up for a year, and after the year ended they told me I have too pay 2000 NIS for the next year (that's 166 NIS a month!) because of high bandwidth usage. So I guess that's what they need the monitoring for. I'm also a happy Netvision customer now. Although having a static IP was pretty cool. Having a static IP in netvision costs me $15/month, signed up for a year. That's around 20 NIS more than not having it. I am not sure about actcom rates, but IIRC they are a bit higher, which makes it even less than 20 NIS more expensive. This is, BTW, for 1.5mbit/s dl. I am not sure about the upload - I think it's 150kbit/s. I did not buy from bezeq higher upload as it costs much more, but I still get around 120kbit/s, although I think my deal says I have 96kbit/s. To buy this from netvision, you have to be a business customer. I do not know what advantages this gives, if at all, except having prices in $ (which IIRC is illegal?). One weird small disadvantage is that you can't change your password from the website - only by phone, only to a new random one. Weird. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Old laptop is firewall, spindown the disk?
On Mon, Sep 19, 2005 at 09:48:53PM +0300, David Harel wrote: Do you know if noflushd will eventually spin the disk up again if cache becomes full? No, the disk will usually spin up automatically when needed. Very old or broken disks might not. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Forcing the use of specific library directories during link and compile
On Tue, Sep 20, 2005 at 12:01:17PM +0300, Shachar Shemesh wrote: David D wrote: Just shooting in the dark: chroot and gcc -L ? Chroot would require me to build another environment, just for the building. No thanks. As was mentioned, there is a tool to do this - debian package sbuild. Never used it myself. Might turn out to be quite comfortable. I think you might want to start by linking the libs you want directly, not using '-l', e.g. gcc -o exec1 f1.o f2.o /path1/lib1.a /path2/lib2.a You might also try 'gcc -v' to see what it does, and also try running collect2 (or ld, which it runs) directly. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Old laptop is firewall, spindown the disk?
On Mon, Sep 19, 2005 at 03:25:59PM +0300, David Harel wrote: Hi, I have this old laptop I turned into a firewall. It is only doing IP forwarding (NAT, MASQUERADING or whatever it takes) using iptables. I have 256MB ram on it. My idea is to have the disk spin down unless it is needed. Any idea? Mount root with noatime. Google for noflushd. Stop all unnecessary daemons. Good luck, -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Linux-based Windows rescue/cleanup cd?
On Sat, Sep 17, 2005 at 12:06:56PM +0200, Hetz Ben Hamo wrote: No, you don't need to buy a seperate CD. The partBE installation uses your machine's XP to create the bootable CD. You cannot install any programs on top of it after you burned it. Heck, you cannot even use internet explorer, or office with it. This is part of Microsoft PE enviroment (Pre-install Enviroment). If It's not, actually. It's similar, but isn't related to PE. you have looked at some newer IBM laptopped and you pressed the Access IBM blue button - you're getting the same thing as BartPE, just with IBM Custom configuration. Thanks, Hetz On 9/17/05, Amit Aronovitch [EMAIL PROTECTED] wrote: Hetz Ben Hamo wrote: I belive in: use the best tool for your job. What I would recommend is not a Linux solution, but sort of mini windows solution. It's called: bartPE, it's GPL'd, and it creates an ISO image from your Why do you think it's GPL? It's not, and won't be, as far as I understand. current Windows XP installation. Prior to making the ISO, you can add few programs (antivirus, ghost, partition magic, firefox, etc) and make the ISO. Then you burn it to a CD-R and boot from it. With this bootable CD, you're getting a mini version of XP and you can access the hard drive and scan for viruses, worms, spyware, adware etc and clean them. URL: http://www.nu2.nu/pebuilder/ Free Windows live CD - is that legal? IANAL, but his page indicates he did invest time on making sure he won't be sued. I suggest you read it thoroughly if interested. I bet you'd have to purchase *at least* one more Windows license (per cd) from MS (most probably it would not be your normal OEM, and would not be cheap). Not necessarily. But don't take my word for this. I am sure this isn't a clearcut. I guess if you use it e.g. for normal work on a lot of machines in a business (e.g. with a single business application for which you can live with the limitations in bartpe), you are taking a quite big risk. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Linux-based Windows rescue/cleanup cd?
On Wed, Sep 14, 2005 at 03:39:09PM +1000, [EMAIL PROTECTED] wrote: Hello, A friend will bring his PC tonight to my place so I can cleanup and try to speedup his Windows XP until his new one arrives. I plan to search-n-destroy viruses and spy-ware and try to make his machine more secure by installing Firefox/Thunderbird and maybe tweak IE/Outlook. His windows is bootable and working but I think it might be better to do some of the cleanup (virus checks, defrag) from a Linux environment. (I don't know which FS type he has). From digging the net so far I found a rave-review of System Rescue CD (http://www.sysresccd.org/) in NewsForge (http://tinyurl.com/9dltu) and the Trinity Rescue Kit (http://trinityhome.org/trk/) is found high in Google search results, but both seem to be ancient (over a year old). Right now I plan to use the System Rescue CD since the NewsForge review describes the situation I expect to tackle tonight. Any pointers/recommendations/warnings are welcome. For quite many years I did not use any bootable CD except for knoppix. I do not know if it has an antivirus but other than that it usually had whatever I needed at the moment. Recent versions even make it quite easy to install stuff from the net to a ramdisk/unionfs. I should also mention a little known fact, that there is a free (as in beer only) tool to create bootable Windows XP CDs. There is a quite big and fruitful comunity around it - I wouldn't be surprized if they already have packaged one of the freely-downloadable AVs. Google for bartpe or pebuilder. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Working for over a year
On Wed, Sep 07, 2005 at 05:45:13PM +0300, Tzafrir Cohen wrote: http://counter.li.org/reports/uptimestats.php Also http://uptime.netcraft.com/up/today/top.avg.html, if you like this kind of literature. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Setting up a private network behind a firewall.
On Mon, Sep 05, 2005 at 02:06:23PM +0300, Gilboa Davara wrote: Hello all, I looking to create a permanent connection between my home network and the private network I have at work. The setup is as follows: Home - Firewall Cable - Internet - ADSL - Win2K - Workstation - Work. I'm currently using the SSH port forwarding to gain access to my private network from work; However, this solution doesn't give me access to my work network from home. I have no control over the Windows 2K gateway/NAT, and if I bother the admins they might revoke my DSL privileges. In short, I cannot ask him to forward specific ports to my machine. More-ever, both IPs ([EMAIL PROTECTED], [EMAIL PROTECTED] are dynamic, though my private network is resolved using freedns.afraid.org) As far as I could see, OpenVPN might solve my problem; but AFAICS it requires both ends to have static IP and/or resolvable hosts. To reiterate: I need a solution in which, my workstation at work, will establish a secure (semi)-persistent connection between my home firewall creating virtual network between the two, giving me access to my work network from home. You might try pppd over an ssh connection. There is a howto about this as well as a debian package secvpn doing this. Never tried it myself. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: devfs - gone
On Mon, Sep 05, 2005 at 04:24:34PM +0300, Shachar Shemesh wrote: Hi all, The new kernel (2.6.13) has removed support for devfs. While I certainly understand why devfs is not on by default, it seems that removing it altogether is a bit harsh. Devfs has been a great way to quickly (say - inside an initrd) get a clear picture of what hardware is available, as well as get the relevant devices required to run said hardware. With devfs gone, installers, some embedded configurations, and possibly others, may find their time much harder. This was discussed quite thoroughly on lkml and summarized on one of the last kerneltraffic issues (which are very recommended BTW), here http://www.kernel-traffic.org/kernel-traffic/kt20050827_318.html My main question is - does anyone know of a good alternative? The recommended alternative is udev. I used it quite successfully, although I know that it's still not a complete replacement. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: OT: Internet connection errors
On Wed, Aug 24, 2005 at 12:09:27AM +0300, Leonid Podolny wrote: I will call both tomorrow with the added data but do not expect much. I'd call Netvision and inform them that you are leaving unless the problem will be solved very soon. They can solve it even if it's Bezeq's Why do you think so? fault. From your point of view, you are a customer that doesn't get the service he pays for. Let them figure it out. You've done far more than they could expect from you. Of course, but I did not do this to help them, but to help me. I think that it's pretty clear that the problem is not at your configuration, so there is hardly anything you can do. Well, I did do some more things :-) I tried, following Netvision's supportperson idea, Bezeq's test connection ([EMAIL PROTECTED], any passwd) and it was the same. So it's clearly not Netvision's fault. I took a modem of the same model from a neighbour, and it happens with it. I also verified that it does not happen with the neighbour's line. Bezeq wanted me to try also pppoe from the linux machine, not using the modem as a router. This also behaved the same. Let's see what their next idea will be. They claim the line is in good quality, according to their monitoring. Thanks, -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
OT: Solved: Internet connection
Hi all, In case anyone was interested: After lots of talks with many mostly-clueless people, they sent a technician, which after 2 seconds of looking at the tons of ping outputs on my xterms, drove a few minutes, moved me to another port (in their (DSLAM?) switch), and [1]ping is now happy. [1] http://ftp.arl.mil/~mike/ping.html -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
OT: Internet connection errors
Hi all, Sorry for posting offtopic. I already talked to Bezeq and my ISP (netvision) and both say there is no problem. Maybe someone here will have an idea. I connect to the internet through Bezeq ADSL+Netvision. The last change in my configuration, at least that I recall, was changing the modem (an ECI 312) to be a router, a few months ago. In the last 1-2 weeks, I had unexplained stops (cuts?) in the connection. After talking today, as I said, with both, I looked a bit further. I ran on the modem itself ping to the other side of the ppp connection (212.143.208.145 currently), and I see that exactly each 519 seconds, the connection goes down for exactly 12 seconds. The ppp connection does manage to stay up - it does not disconnect/reconnect. But ping does not reply for 12 seconds. I now discovered that the connection dies only on one direction - from the net to me. I had a gnomemeeting talk with someone and he said that during the cut he did continue hearing me. Any idea anyone? Thanks. I will call both tomorrow with the added data but do not expect much. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: xterm/mlterm
On Sun, Aug 21, 2005 at 12:05:01PM +0300, Avraham Rosenberg wrote: Hi, Your problem with the mlterm settings, which apparently was not addressed in the answers you received, caught my eye, because I am also contemplating moving to UTF8, and because I grew to be addicted to my habits. I would, therefore, like to hear from you how these fare in mlterm. I still did not play with it much, but I am pretty sure I don't and won't like it. I use it currently almost only for email, with most of my windows being xterm. Few of the things I miss: Double-click behaviour is different - dragging doesn't select by words, only the first selection is a whole word It's a bit less flexible in setting word boundries No extend selection (right click in xterm) Most important (and I really would like to know what sets this option) is the interpretation of the modifier key: I like to use it in the shell like it is defined in emacs: I mean ALT-F, ALT-b to jump a word forward/backward, ALT-D, ALT-BS to erase one, ALT-c to capitalize, ALT-u to change to upper case till the end of the word and ^ALT-y in conjunction with ALT-i to recall the i-th argument of the preceding command. Did not play with this. I use vi :-) For the rest of my preferences, I quote the relevant lines from .Xresources: XTerm*title: Hebfont xterm XTerm*geometry: 100x39+0+0 XTerm*scrollBar: on XTerm*saveLines: 500 XTerm*pointerShape: hand2 I think most of these aren't an issue (some I also tried). XTerm*background: cornsilk XTerm*foreground: black I personally like dark backgrounds. This is what I use: #!/bin/sh bg=rgbi:`dd if=/dev/urandom bs=1 count=3 2/dev/null | od -tu1 | \ awk '{b=0.1/256 ; printf(%s/%s/%s\n, $2*b, $3*b, $4*b); exit(0)}'` exec xterm -bg $bg -fg white $@ and it's the exact same in mlterm (-fg/-bg). XTerm*visualBell: false XTerm*VT100*font: -*-*-*-*-*-*-24-*-*-*-*-*-iso8859-8 Which font is this in reality on your machine? I have personally edited the Hebrew part of the utf-8 10x20 and use it. You can get it at http://www.cs.tau.ac.il/~didi/10x20heb.tgz. It will probably be a bit small for you if you use 24. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: OT: Reversing the linker's action
On Thu, Aug 18, 2005 at 09:59:26AM +0300, Efraim Yawitz wrote: On Thu, 18 Aug 2005, guy keren wrote: the process is not reversible, because 'linking' is not a reversible function - it only takes part of the objects and places them in the binary (especially if we're talking about objects taken from static libraries). also, (here i am half-guessing) this object information doesn't have to be in the resulting executeable at all. further, there could be functions that got lost on the way. btw, if you're trying to do this in order to comply with some LGPL license - there are better ways (dynamic linking). I just want to know how to take a binary which was dynamically linked with a newer version of glibc than I have and turn it into a statically linked binary that I can run on my system. Search for a project called statifier. Never used it myself. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: cpp compile problems ? (Debian)
On Wed, Aug 17, 2005 at 04:43:05PM +0300, Peter wrote: My g++ compiler seems to be misconfigured. It is gcc/g++ 3.3.3 from Debian and it cannot find simple things, like cout. Example: /usr/include/c++/3.3/iostream contains: extern ostream cout; My test program is: #include iostream int main( int argc, char *argv[] ) { cout Hello, World!\n; exit(0); } And compiling yields: [EMAIL PROTECTED]:~/download/c++-check$ make 1 g++ 1.cpp -o 1 1.cpp: In function `int main(int, char**)': 1.cpp:5: error: `cout' undeclared (first use this function) 1.cpp:5: error: (Each undeclared identifier is reported only once for each function it appears in.) make: *** [1] Error 1 HUH ? Any input will be welcome, thanks, Either add 'using namespace std' or #include the older iostream.h. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
bidi xterm
Hi all, This issue comes up every now and then. Now is the next such time :-) I use and love xterm. I want bidi. I use for several months mlterm, which is ok. But some of the differences annoy me. It might be possible to conf mlterm to behave more like my (tuned) behaviour of xterm but I did not try that (yet?). There was a patch by Rebert Brady, which I can't find anymore on the web. Is it this one? http://www.debian.or.jp/~kubota/softwares/xterm-152-27.diff.gz pointed at from http://www.debian.or.jp/~kubota/xterm.html ? It clearly does not apply cleanly against latest (xterm-204) version, as it's 4 years old. Well, anyone using this patch? What should be done (if at all) to get it integrated into xorg/Dickey's? -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Open source driver Fiber Channel devices for Linux?
On Tue, Aug 09, 2005 at 04:26:44PM +0300, Shachar Shemesh wrote: Hi all, Can anyone recommend to me a Fiber channel storage device (HBA) that has open source drivers in Linux? I am being offered something called FC2-133 (a.k.a 24p0960). Some initial research shows that the Linux drivers for this adapter are binary only, which means I'm looking into a series of kernel compilations, much prayer, and things not working. It also means I'm locked to one or two specific distros. I would rather not be in this mess. I'm wondering what are other people doing in that respect? What hardware do you use? I saw that a company called Emulex is producing open source drivers for kernel 2.6 series for their HBAs. These do not appear to be part of the actual vanilla kernel, but are shipped with Suse. I'm wondering whether anyone has any experience with that? We use Emulex lp1. There is a driver for it, named lpfc, built into recent 2.6 kernels. We use RHEL3 (2.4) which also has it. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Advice - fastest INTEGER operations CPU
On Sun, Aug 07, 2005 at 02:52:31PM +0300, Shachar Shemesh wrote: Hi all, I'm looking into buying a computation server for a client. They are looking for the platform that will give them optimal INTEGER performance. I'm thinking between the 64Bits - PowerPC, Itanium and the EMT64/AMD64 technologies. I am also interested in more specific knowledge (Xeon is better than Athelon etc.). Thoughts? Ideas? There are 3 kinds of lies: Lies, damn lies, and benchmarks. If at all possible, run your client's app on all relevant platforms. That said, you might want to look at spec.org - they have specific benchmarks. E.g. SpecInt2000 for integer cpu speed. Do note that there are there quite noticable differences between machines with the same CPU(s), which means that even for raw cpu speed, not all machines are created equal. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: low-level formatting?
On Tue, Aug 02, 2005 at 09:01:55AM +1000, Amos Shapira wrote: On 8/1/05, Karasik, Vitaly [EMAIL PROTECTED] wrote: According to e2fsck manpage, you should run fsck -cc (man e2fsck) for marking bad blocks. I haven't tried this. That's not low-level format - what it does is to tell the filesystem code how to avoid bad blocks. But if I want to install, e.g. a swap partition or Windows then I might be out of luck, as there is no way to tell them to avoid the bad blocks (that I'm aware of). mkswap also has '-c'. In windows I think that doing a full (not quick) format will also find and mark bad blocks. I would still be interested in an answer to your question. I know (and used) such tools for SCSI disks, not IDE. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: need script to find alien files
On Wed, Jul 27, 2005 at 05:12:53PM +0300, Michael Green wrote: I'm looking for script that will traverse filesystem of an RPM-based distro and find files that do not belong to any RPM. I do not know about such a thing, but I happend to do a few times something like rpm -qal | sort file1 find / | sort file2 Then diff file1 file2. Of course, the diff will be very big. Some of it you can skip/delete automatically - e.g. /dev (if you have devfs or udev). A good script will do this for you, but even without one, it's not that hard. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: chumash for lin?
On Thu, Jul 28, 2005 at 08:17:54AM +0300, Aaron wrote: Hi all, I was wondering if any classic Jewish texts, chumash mishnah etc are available for linux. I am looking for study material for my daughter who has a test in chumash. Has anyone written an application for torah study on linux? I don't know about Hebrew ones. There are few English ones - e.g. search freshmeat for bible or apt-cache search bible. Some of them might work with Hebrew texts, which you can find on the net (e.g. at mechon mamreh). Adapting the texts to work with English software might be non-trivial. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: system clock loops
On Mon, Jul 25, 2005 at 09:06:42PM +0300, Tzafrir Cohen wrote: On Mon, Jul 25, 2005 at 06:19:02PM +0300, Ehud Karni wrote: On Mon, 25 Jul 2005 00:36:42 +0300, Amit Aronovitch wrote: Geoffrey S. Mendelson wrote: I am a Nevtvision customer, they use a GPS unit to give them a stratum one server. I would avoid the overloaded server at HUJI if at all possible. Netvision here too - but I see them as stratum 2. I sync to ntp.netvision.net.il (+ 2/3 times europe.pool.ntp.org for backup). Do they have another address for the GPS machine? You can use ntp.ilan.net.il (aka ntp.net.il) - startum 1 (using lab atomic clocks - not GPS - I think it is more accurate). Isn't that the overloaded HUJI server? Stratum 2 Israeli ntp servers: timeserver.iix.net.il , openu.ilan.net.il . And the netvision server. All seem to sync from that startum 1 server at HUJI. No, timeserver.iix.net.il has its own gps. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: using Palm to transfer PC files
On Fri, Jul 15, 2005 at 02:03:30PM +0300, Shlomo Solomon wrote: Hi, I'd like to be able to copy PC files to and from my Palm. I've googled and haven't found this. I know something like this exists for Windows. Does anyone know of a LINUX equivalent? pilot-schlep from pilot-link. Never used it myself. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: my new Palm Zire 72
On Sat, Jul 09, 2005 at 09:31:53PM +0300, Shlomo Solomon wrote: (Snipped Bluetooth question - I never used Bluetooth) I already read your later message about 95% success. So :-). am also pretty sure it's a (partly) physical problem of the connection, not (only) a software one. I doubt that. I've tried different USB plugs (back panel, front panel and via a Hub) and 2 different cables. I think I already wrote that I have no problems with my printer, scanner, web-cam, disk-on-key or mouse (all USB). I did not tend to imply you had a hardware problem, only said I think I did. Hardware problems are easier to solve by cross-checking, as you did, and I assume you came to the correct conclusion that it's not hardware. I did not thoroughly read all your tests and results. I do have two points to make, some of them I already said in earlier posts. 1. There is no magic in /dev/pilot. The hotplug scripts choose the Although on the tests I sent yesterday I tried /dev/pilot, I do know that this is only a pointer and might not point correctly. So I did do several test directly to the ttyUSB* devices and had no better results. In fact, it seems to me that /dev/pilot IS being correctly defined. Look at the following test: - before connecting the cable --- [EMAIL PROTECTED] solomon]$ ls -la /dev/pilot ls: /dev/pilot: No such file or directory [EMAIL PROTECTED] solomon]$ ls -la /dev/ttyU* ls: /dev/ttyU*: No such file or directory connect the cable -- [EMAIL PROTECTED] solomon]$ ls -la /dev/pilot lrwxrwxrwx 1 root root 12 Jul 9 21:14 /dev/pilot - /dev/ttyUSB1 [EMAIL PROTECTED] solomon]$ ls -la /dev/ttyU* crw-rw 1 solomon uucp 188, 0 Jul 9 21:14 /dev/ttyUSB0 crw-rw 1 solomon uucp 188, 1 Jul 9 21:14 /dev/ttyUSB1 [EMAIL PROTECTED] solomon]$ pilot-xfer -p/dev/ttyUSB0 -l Listening to port: /dev/ttyUSB0 Please press the HotSync button now... Ctrl-C to stop this -- [EMAIL PROTECTED] solomon]$ pilot-xfer -p/dev/ttyUSB1 -l Listening to port: /dev/ttyUSB1 Please press the HotSync button now... Error accepting data on /dev/ttyUSB1 [EMAIL PROTECTED] solomon]$ pilot-xfer -p/dev/pilot -l Listening to port: /dev/pilot Please press the HotSync button now... Error accepting data on /dev/pilot Notice that both /dev/pilot and /dev/ttyUSB1 give the data error message (not surprising since they're really the same device). I interpret this to mean that some attempt to communicate is being made and failing while the USB0 device gets no reaction. Am I wrong? You did not say if these messages are before pressing hotsync, or after, or both. Can you try again both before and after, on all 4 devices? My guess is that connecting the cable creates two devices (0 and 1) that do not work, and pressing hotsync creates another two (2 and 3), and one of them works. I guess my guess isn't accurate, but I can't check it. I did see, though, a USB-to-Palm cable that was also a simple card reader (only for the palm's cards). The card reader was active only if you toggled a switch on it, and was (obviously) active on connection, without pressing hotsync. Maybe yours is somewhat similar? After writing the above and (obviously) before sending this message, I succeeded once more - using /dev/pilot - so I do think /dev/pilot IS pointing to the correct port (ttyUSB1). But after the one success, again no luck :-( Did you check that when it worked, /dev/pilot really linked to USB1? You could try dlpsh instead of pilot-xfer if you want time for checks (it finishes the connection only when you tell it). 2. The behaviour you describe is definitely different from what I see here (with all 3 devices) - none of them cause the creation of any /dev/ttyUSB device on connection, and all cause creation of 2 devices (0 and 1 if it's the only device connected) when pressing hotsync in the palm. They differ in which of the two devices actually work. again - I do think /dev/pilot is being created correctly, but I have tried all the other created devices. Maybe. I don't know. 95% is good, but not enough. The two tungstens here connect few times a day for few months, and rarely have problems. Or at least the rarely tell me they have (they aren't mine). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: my new Palm Zire 72
On Fri, Jul 08, 2005 at 02:36:59PM +0300, Shlomo Solomon wrote: On Friday 08 July 2005 10:05, Shlomo Solomon wrote: I now intend to clean everything up and make a few more tries. I'll try to keep acurate records of what I do and see if the log entries can be of any help. OK - I did several experiments. Unfortunately, I was unable to reproduce the situation I mentioned befor when pilot-xfer worked 2 times in about 20 tries. But here are the results of my experiments. I tried 3 different USB sockets on my machine. Below I've written what I did each time and included syslog entries. I am very happy about your progress. I started feeling really bad about this. For the record - I did see differences between the connection reliability of different palms connected to the same usb cable. Tungsten T3 was more problematic than Tungsten T and m130. But the problems are occasional, not systematic or as frequent as yours. I am also pretty sure it's a (partly) physical problem of the connection, not (only) a software one. There is no point in doing Ctrl-Z. If you want to shoot - shoot, don't talk. Ctrl-C. I did not thoroughly read all your tests and results. I do have two points to make, some of them I already said in earlier posts. 1. There is no magic in /dev/pilot. The hotplug scripts choose the device they think is the right one and make /dev/pilot a link to it. It's very possible that they are wrong - as I said, it took me a lot of work to automatically make only the above 3 devices work, and the hotplug scripts intend to support theoretically all the devices. So, when you return to playing with this, do the following: connect the device/cable/hub etc. Press hotsync. Then try pilot-xfer or whatever with /dev/ttyUSB[0123] directly, not /dev/pilot. Each time try another one. I am pretty sure one (and probably only one) will work, and will work all the time. 2. The behaviour you describe is definitely different from what I see here (with all 3 devices) - none of them cause the creation of any /dev/ttyUSB device on connection, and all cause creation of 2 devices (0 and 1 if it's the only device connected) when pressing hotsync in the palm. They differ in which of the two devices actually work. I never tried connecting through a hub, as far as I recall. I do not think a hub should matter, assuming it's otherwise working well. Good luck, -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: threads in 2.6
Replying to myself ... On Wed, Jul 06, 2005 at 01:16:46PM +0300, didi wrote: Hi all, If a process creates threads, then the main thread dies, I can't see the other threads anymore with ps. I also do not see them with 'ls -l /proc'. If I do know the pid (tid actually), and do ls -l /proc/tid, I do see it. Is this intended, or a bug? In 2.4 I always see all of the threads. Tried on 2.6.12.2. I spent some more time on this and have some updates: It's a known bug. E.g. http://www.ussg.iu.edu/hypermail/linux/kernel/0410.0/1589.html I tried to debug it. I have quite little experience with kernel programming, so after making a small change that did not work, I decided to try user-mode-linux for this. Under UML, a small test program I wrote crashes with SIGSEGV some time after calling clone. I am pretty sure it dies on the first context switch between the two threads, or something like that - not on a syscall (and I do not think that on a memory access of mine - to test I wrote trivial busy loops). The test program does work on real 2.6, with the behaviour described above. I thought this might be a bug in libpthreads, so tried to write it with direct calling of clone, and did not manage to make it work, I don't know why. Various versions of the test program can be found in http://www.cs.tau.ac.il/~didi/threads. The solution I hope to find/write will cause a zombie process keep a working /proc/tgid/task subdir with subdirs of its threads, living ones still accessible as normal. I am pretty sure (without checking) that ps will magically start showing them without changes. I guess the bug can be solved somewhere around fs/proc/base.c - maybe in proc_task_lookup. I will try to make some changes, but this obviously takes much more time than with uml. If I see that it is too slow I might try XEN. I gave it a quick look and see there are problems with tls libs under xen, so it might not help me debug this. I hope I do not bother most people too much - I know it's offtopic, I know there are people here that might be interested in solving this, or at least that it's solved. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: my new Palm Zire 72 - resend
On Fri, Jul 08, 2005 at 12:02:12AM +0300, Shlomo Solomon wrote: strange - I sent 2 messages to the list, but only the 2nd one got througt - so here's the 1st one again. On Tuesday 05 July 2005 22:59, Shlomo Solomon wrote: On Tuesday 05 July 2005 19:52, Yedidyah Bar-David wrote: OK. Let's start from the very beginning. First, start from a clean known state. Either after a reboot, or try to For completely unrelated reasons, re-booting is not an option today. I hope I'll be able to do the tests you recommended in a day or two and I'll let you know what happens - thanks. On thinking it over again, and after a few more of my own experiments, I can't see how my problem will be solved by re-booting. After all, that's not really the LINUX way, is it :-) Can you at least rmmod visor and usb_serial (and modprobe them if needed), then repeat the tests? -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
threads in 2.6
Hi all, If a process creates threads, then the main thread dies, I can't see the other threads anymore with ps. I also do not see them with 'ls -l /proc'. If I do know the pid (tid actually), and do ls -l /proc/tid, I do see it. Is this intended, or a bug? In 2.4 I always see all of the threads. Tried on 2.6.12.2. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: my new Palm Zire 72
On Tue, Jul 05, 2005 at 06:23:30PM +0300, Shlomo Solomon wrote: [snip] I still think it's strange that plugging in the USB cable causes the Kpilot icon to pop up, so something is, at least partly, set up properly. Not necessarily. Something is set up to respond to USB hotplug events. Maybe not properly. OK - let me clarify somethig. USB4 and 5 was just an example. Each time I try, 2 new USB devices get created (and sometimes get deleted after a while). At the moment, here's how my system looks (with 12 devices): [snip] OK. Let's start from the very beginning. First, start from a clean known state. Either after a reboot, or try to rmmod both visor and usb-serial. Make sure you have no /dev/ttyUSB*. Then do 'dmesg -c' to clean the kernel's buffer (maybe into some file if you want to keep it). Then do the following things. After each of them, do 'dmesg -c' into a new file. This will let us see what the kernel says at each point. 1. Plug in your cable. 2. Connect the palm to it. 3. Press the hotsync button. 4. Wait until the palm times out. 5. Disconnect the palm from the cable. 6. Unplug the cable. Now do again all of 1-6 (to see what happens on a second time). I hope that the first time will only show USB0 and USB1, as it does for me. Maybe something in kde or kpilot keeps the device(s) open, and that's why on subsequent tries you get new devices. To try that, do all of this without KDE or something smart like it (try e.g. fvwm, wmaker etc. or even a text console if you feel comfortable enough in it). Now, assuming it does start with USB0 and 1, start again from a clean state, then, after you press the hotsync button, try pilot-xfer. First on USB0, then 1. Tell us what happend. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: users' .bashrc not getting executed?
On Mon, Jul 04, 2005 at 09:50:46AM +0300, Ira Abramov wrote: I thought the new question needs a new thread. any answers appreciated! * On a bunch of RHEL 3 machines which had winbind added; perhaps because of the pam games, the .bashrc are not executed (not after su -, nor when you ssh into the machine). Anyone care to point me in the right direction? the permissions on the bashrc are correct, but only root gets his .bashrc executed. users on the local passwd and from winbind don't. bashrc isn't sourced on login shells, only on interactive non-login shells. I guess root's .profile or .bash_profile sources .bashrc or something like that. If you are certain that it started after adding winbind, then I have no idea. Maybe winbind/your pam changes cause the shell to be non-login? I must admit your email caused me to spend a few minutes debugging an annoying problem I have for a few years. I have (in both .bashrc and profile) 'source ~/.bash_aliases'. In .bash_aliases I have alias a=alias and then many lines like a j=jobs On interactive shells (both login and non-login), it all works well. But when I scp to this machine/account, I get many lines like /home/didi/.bash_aliases: line 5: a: command not found Since it causes no harm, I did not check why it's so until now. Now I did, and after another look in 'man bash', I see Aliases are not expanded when the shell is not interactive, unless the expand_aliases shell option is set using shopt -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: users' .bashrc not getting executed?
On Mon, Jul 04, 2005 at 11:00:57AM +0300, Nadav Har'El wrote: On Mon, Jul 04, 2005, Yedidyah Bar-David wrote about Re: users' .bashrc not getting executed?: bashrc isn't sourced on login shells, only on interactive non-login shells. I guess root's .profile or .bash_profile sources .bashrc or something like that. I wonder why this is so. The situation is certainly differrent in Ksh and in Zsh, where the $ENV or .zshrc is executed on *any* interactive shell, login shell or not. The traditional division between the profile and rc ($ENV in Ksh nomenclature), was that the profile was read only in one login shell that is the parent of all others, and thus should contain only inheritable settings, especially exported variables and umask settings. The rc file would contain things that cannot be inherited and are specific to interactive shells (on the last point, the behavior in tcsh is different), such as aliases, functions (which were inheritable in the original ksh, by the way), and stty settings. But what if ENV isn't set in profile? Nothing gets sourced (in sh-compatibility mode). ash doesn't even have a compatibility mode - that's its only behaviour (only /etc/profile ~/.profile and ENV if set). So I was suprised to learn today that indeed, bash default behavior is not to run the .bashrc on an interactive login shell. I find this extremely strange... By the way, I see now in bash's manual, that if it is invoked as sh (rather than bash), it uses the ksh-like ENV variable and its behavior. I agree it's a bit strange, but it's not a real problem practically. Only causes bugs such as Ira's (if that indeed was the thing). I personally simply have .profile linked to .bashrc and that's it. P.S. I'm a very happy user of Zsh. Yes, we know that. You remind us often :-) I admit your pressure was enough that I tried it a few years ago, after using tcsh for many years, for a few months, and then moved to bash. Of course zsh is more powerful, but bash is more standard, so I can rely on it being installed and being the default in almost all new machines I encounter. It has its share of peculiarities, but not this one, so you can consider switching :-) Indeed a good enough reason :-) -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: my new Palm Zire 72
On Mon, Jul 04, 2005 at 09:07:02PM +0300, Shlomo Solomon wrote: This should be easy, but for some reason, I can't seem to sync my new USB Palm. My previous Palm was a serial one and pilot-xfer -p/dev/ttyS1 worked fine. When I plugged in the USB cable on my new Zire 72, a KPILOT icon popped up on the KDE desktop so I thought - wow, this is going to be easy. But, although the icon popped up (and usbview also recognizes the Palm), it doesn't work. I tried autodetection in the Kpilot configuration wizard, but it seems to be looking for /dev/pilot which doesn't exist. I looked in the /dev directory and didn't find anything. USB Palms use either /dev/ttyUSB0 or /dev/ttyUSB1, depending on model. Use e.g. something like 'dlpsh -p /dev/ttyUSB0' (from pilot-link) to find out which one, and make /dev/pilot a link to it. If someone could walk me through the setup, it would be greatly appreciated. Feel free to contact me if you need help. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: my new Palm Zire 72
On Mon, Jul 04, 2005 at 10:56:55PM +0300, Shlomo Solomon wrote: On Monday 04 July 2005 21:46, Yedidyah Bar-David wrote: USB Palms use either /dev/ttyUSB0 or /dev/ttyUSB1, depending on model. Use e.g. something like 'dlpsh -p /dev/ttyUSB0' (from pilot-link) to find out which one, and make /dev/pilot a link to it. I tried the dlpsh command (even though neither of these devices exist), but as I expected there was no result. OK. For a start, try to create it manually. # mknod /dev/ttyUSB0 c 188 0 # mknod /dev/ttyUSB1 c 188 1 If it works, you might try playing with udev etc. if you want. If not, maybe you need to manually load the module - I don't know if hotplug does that automatically (and you did not say if you use hotplug but I guess you do). # modprobe visor Ohhh, one more thing - maybe that's the problem: The device is connected only when you do hotsync (with the cradle or the hotsync app). Try to press it and then check stuff (if you have a device etc.). Otherwise it does not appear to be connected (e.g. you won't see it in lsusb). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: my new Palm Zire 72
On Tue, Jul 05, 2005 at 01:32:33AM +0300, Shlomo Solomon wrote: OK - I think I'm making some progress here. Each time I connect or At last :-) disconnect the USB cable, there are changes in the /dev directory. Notice that there are several USB devices being created - always two at a time. I'm also including dmesg output below. But I've tried using each of the USB devices to sync and I get no response from the Palm. For example, the command pilot-xfer -p/dev/ttyUSB5 -l and running hotsync does nothing and eventually, the Palm hotsync application times out. [EMAIL PROTECTED] solomon]$ ls -la /dev/ttyUSB* crw-rw 1 solomon uucp 188, 0 Jul 4 22:33 /dev/ttyUSB0 crw-rw 1 solomon uucp 188, 1 Jul 4 22:33 /dev/ttyUSB1 crw-rw 1 solomon uucp 188, 2 Jul 5 00:54 /dev/ttyUSB2 crw-rw 1 solomon uucp 188, 3 Jul 5 00:54 /dev/ttyUSB3 crw-rw 1 solomon uucp 188, 4 Jul 5 00:55 /dev/ttyUSB4 crw-rw 1 solomon uucp 188, 5 Jul 5 00:55 /dev/ttyUSB5 [snip] -- from dmesg usb 2-2: new full speed USB device using address 54 visor 2-2:1.0: Handspring Visor / Palm OS converter detected usb 2-2: Handspring Visor / Palm OS converter now attached to ttyUSB4 usb 2-2: Handspring Visor / Palm OS converter now attached to ttyUSB5 [snip] OK. I have no idea about how to continue. I'll just summarize my experience. I never used a Zire72. I did use (i.e. connected to a linux machine) a VX (serial cradle), and m130, Tungsten T, Tungsten T3, Zire31 (only a few times). All of them, IIRC, and at least the Tungstens for sure, emit two connections in dmesg - e.g. Jul 3 08:01:59 maint kernel: usb 1-1: Handspring Visor / Palm OS converter now attached to ttyUSB0 Jul 3 08:01:59 maint kernel: usb 1-1: Handspring Visor / Palm OS converter now attached to ttyUSB1 Jul 3 08:02:03 maint kernel: visor ttyUSB0: Handspring Visor / Palm OS converter now disconnected from ttyUSB0 Jul 3 08:02:03 maint kernel: visor ttyUSB1: Handspring Visor / Palm OS converter now disconnected from ttyUSB1 But, as I said, some are actually accessible from ttyUSB0 and some from ttyUSB1. I wanted to find out automatically which one, which wasn't easy (found no real info on google). So I simply tried, and at least for the first 3, I use the following script: #!/bin/sh MAX=30 # Fallback dev=ttyUSB1 tmp1=`mktemp /tmp/get-palm-dev.XX` n=0 while [ $n -lt $MAX ]; do lsusb -v $tmp1 if cat $tmp1 | awk '/Palm/ {palm=1; p=$0} palm /bcdUSB/ {print p, $0; exit}' | grep -q 'P alm Tungsten T.*bcdUSB.*1\.00'; then dev=ttyUSB1 break elif cat $tmp1 | grep -q Palm; then dev=ttyUSB0 break fi sleep 1 echo -n . 12 n=`expr $n + 1` done rm $tmp1 echo $dev That is, the only difference between the Tungstens is the bcdUSB, which is IIRC 1.10 with the T3 and 1.00 with T (or vice-verse). I do not think it will work for other palms without tweaking, so do not use it as is. I never saw one that used 4 and 5 like yours. A first guess would be that you also have other usb-serial hardware connected, but you say that before connecting there are no devices, which rules it out. Did you try also 0-3 (Even though the kernel says 4 and 5)? Doesn't google say anything useful about Zire72? You might want to look at linux/Documentation/usb/usb-serial.txt. Are you sure the hardware is ok? Does it work in Windows? You might even be able to find out what device it uses in Windows (I have no idea how). BTW, which kernel version? Did you try doing this as root (not needed here, but lsusb -vv does need it on some combinations of kernel/filesystem (there are both usbfs and the older usbdevfs)? -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Webcam
On Sat, Jul 02, 2005 at 12:17:34PM +0300, Matan Ziv-Av wrote: On Fri, 1 Jul 2005, Diego Iastrubni wrote: How about something cheaper? Does any low end (dirt cheap as you called) camera have linux support? There is the qcam 310, which includes a microphone (analog), and can be found for 60 NIS on zap. It works with the spca5xx driver, which does have some quirks, but is in active development. For the record: I have a Logitech QuickCam Express (046d:0920, there are other models of the same name), and it works with spca5xx. The driver does not support compression (yet?) on this (and many others) webcam, and so it's quite slow - around 4 FPS (it's USB 1.1). The image quality isn't bad, if you only use it to talk to friends. You can see here snapshots I just took: http://www.cs.tau.ac.il/~didi/qcsnap1.png http://www.cs.tau.ac.il/~didi/qcsnap2.png http://www.cs.tau.ac.il/~didi/qcsnap3.png I never used myself the 4000, but I also heard that it's much better. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Webcam
On Sat, Jul 02, 2005 at 12:30:39PM +0200, Hetz Ben Hamo wrote: Didi, I'm not sure, but I think the compression which is used with your qcam is being used in the QuickCam 4000, so if anyone wants to tinker with the driver and add the compression support that the 4000 has, would might have a nice quality of picture.. As I said, I never used the '4000. I do not know if its compression is known to linux drivers. I know people did work on this compression. I do not know how much they did. I am sure it's not easy to reverse-engineer. But I did not talk about speed/compression, but about optical quality. The 4000 supposedly gives much better pictures. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: noip service on a b-focus 312
On Thu, Jun 30, 2005 at 12:07:41AM +0200, Lior Kesos wrote: I put the noip binary on a web server on my host and then ran wget from the router while I was on /var that seems to be writeable. Just note that /var is tmpfs, recreated at reboot. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Directing kernel messages to file
On Wed, Jun 22, 2005 at 02:49:38PM +0300, Shachar Shemesh wrote: Hi all, I need to direct the kernel messages to a file, instead of going to a tty. I know how to direct it to a serial console, but I want it not displayed on any interactive medium at all. Is it at all possible? I don't think so. Simply disable them (with dmesg or quiet=) and log them with klog/syslogd. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: slashdot: m$ praises unix and linux shell clis and prepares to emulate them
On Fri, Jun 10, 2005 at 01:05:30PM +0300, Oron Peled wrote: MS people were really puzzled (CLI must be an old art that should die when people get accustomed to the modern GUI world, isn't it?) They took notice and said W2K would be a different story... Yeh, sure. Well, yet another defendant for MS (what happens here, people?). I am not sure about a different story, but W2K's cmd.exe is in some ways better than NT4. At least one thing I noticed is that it addopted most of the redirection syntax and semantics of /bin/sh. E.g. '21' works. Just in case someone here needs that - I know I used it once or twice. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Accessing eci b-focus 312+ ADSL Router/Modem
On Fri, Jun 03, 2005 at 06:58:37PM -0300, Roberto S. Meyer wrote: They often use 10.0.0.1, 192.168.1.1, 192.168.100.1, c. The two 312's I played with had 10.0.0.138, the same as the Alcatel STH (and contrary to ECI 270, which had IIRC 192.168.1.1). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Mess with additional gcc
On Thu, May 26, 2005 at 12:02:38PM +0300, Boris Gorelik wrote: Hello, list. Can anybody help me with this issue? I need to install gcc3.2.2 on my Mandriva 2005, whithout removing the existing gcc3.4 . So I do [builddir]$ sorcedir/configure --prefix=/opt/gcc322 --program-suffix-3.2.2 No need for --program-suffix-3.2.2 if you change --prefix. It will be less convinient if you do. [builddir]$ make [builddir]$ sudo make install Everything works smoothly, but after the install is done, lots of programs fail to start: [tmp]$ konsole konsole: //opt/gcc322/lib/libgcc_s.so.1: version `GCC_3.3' not found (required by /usr/lib/libstdc++.so.6) (the error message is always the same). Removing /opt/gcc322 solves the problem. Removing /opt/gcc322 from where? From PATH or from LD_LIBRARY_PATH? Generally, I do not think it's recommended to run programs with shared libs that are compiled with different compilers. So use your gcc322 (in LD_LIBRARY_PATH) only for programs compiled with it. I know for sure that some versions are known to be incompatible. IIRC the big jump was exactly between 3.2 and 3.3, but I might be wrong. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Mess with additional gcc
On Thu, May 26, 2005 at 01:55:50PM +0300, Boris Gorelik wrote: On Thu May 26 2005 12:29, Yedidyah Bar-David wrote: Removing /opt/gcc322 from where? From PATH or from LD_LIBRARY_PATH? removing as in rm -fr (or mv gcc322 /tmp/gcc322.backup) OK. You did not state exactly what you did, so I'll have to guess. You compiled stuff, using shared libs from this new gcc. You somehow made the execs find those libs (e.g. LD_LIBRARY_PATH or -rpath). Then you deleted it, and they can't find the shared libs. You can't solve the problem without this removed gcc. Do you have backups? You can compile and install gcc322 again, and hope you got the exact same binaries. This will probably work, but there were in the past versions of gcc that were dependant on the compilation environment. So you theoretically might get different binaries if you changed something in your system (e.g. gcc/libc/binutils versions). Assuming you did restore or recompile gcc322, your binaries will work. If you can also recompile the binaries, you can statically-link them so that they do not need the shared libs. If you can't, you can try a program called statifier (search freshmeat), which I never used, which creates a static exec from a dynamic one+libs. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Mess with additional gcc
On Thu, May 26, 2005 at 03:43:17PM +0300, Boris Gorelik wrote: On Thu May 26 2005 14:44, Yedidyah Bar-David wrote: You compiled stuff, using shared libs from this new gcc. You somehow made the execs find those libs (e.g. LD_LIBRARY_PATH or -rpath). Then you deleted it, and they can't find the shared libs. Sorry. This is some more info: There is only one app that is compiled with gcc322. I write this app, since I use a third party library (static linking), and it's headers files do not compile with gcc3.4 (huge mess with templates) . The rest of the apps are standard ones that come with the distro (Mandriva). When the two versions of the compiler are installed, I experience the problems described in the original message. Removing gcc322 (rm -fr gcc322) solves everything, except for the fact that I cannot continue working on my program (which is VERY important). [tmp]$ env | grep gcc322 | wc -l 0 [tmp]$ I didn't create any symlink to any file in /opt/gcc322 Hope this clarifies the situation Not enough. You have a standard install of mandriva. In addition, you compiled gcc322 with --prefix=pref1. Up until now, nothing should have changed. Standard binaries continue to use the standard shared libs, and if you compile an app it will use the standard compiler and shared libs. Now you want to compile app1 with gcc322. How do you do this? First, you need to get to the compiler. You can either run it with the full path, or add pref1/bin to your path. Any of these will allow compiling, but not running, and standard binaries should not be affected. To run app1, you also need to let it find gcc322 shared libs. You have two normal options: either at link time (-rpath) or at runtime (LD_LIBRARY_PATH). If you choose the latter, you also affect the standard apps, that will stop running (because they can't run with gcc322 libs). If you then rm pref1, app1 won't find its libs anymore, but standard apps will resume working (as ld.so can't find libs there, and will use standard ones). Am I describing what happaned? If yes, then you have two options: The easy one, is to not change path and LD_LIBRARY_PATH globally, but use an alias for that. Only in shells you run this alias, you can compile/run app1. This won't work if you need to run both app1 and standard apps from the same shell. So you can take the harder one - add to the gcc link command -Xlinker -rpath pref1/lib. It's not that harder if you run it yourself, but if its source downloaded from the net it will require changing some of its files, as opposed to only changing an envvar. Then this exec will run also from shells not defining LD_LIBRARY_PATH. Now, if this does not describe your problem, please give more details. How do you compile/run your app, env vars you changed, etc. Maybe you added pref1 to /etc/ld.so.conf and did not tell us? BTW, all this refers to C++ - C does not have gcc-specific shared libs, only glibc. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Mess with additional gcc
On Thu, May 26, 2005 at 07:07:30PM +0300, Boris Gorelik wrote: On Thu May 26 2005 16:42, you wrote: On Thu, May 26, 2005 at 03:43:17PM +0300, Boris Gorelik wrote: Not enough. You have a standard install of mandriva. In addition, you compiled gcc322 with --prefix=pref1. Correct. Up until now, nothing should have changed. Standard binaries continue to use the standard shared libs, and if you compile an app it will use Theoreticaly, yes. But something went wrong, and it is broken now. Again, I didn't touch any environmental variable (LD_LIBRARY_PATH, PATH, etc). the standard compiler and shared libs. Now you want to compile app1 with gcc322. How do you do this? First, you need to get to the compiler. You can either run it with the full path, This is exactly what I do: CC=/opt/gcc322/bin/g++322 or add pref1/bin to your path. Any of these will allow compiling, but not running, and standard binaries should not be affected. To run app1, Again, I need gcc322 only for one app, which is linked statically, so passing to the linker appropriate -L and -l parameters do the job. I know that this sounds strange and funny, so I'll stress it again: installing gcc322 in completely isolated directory (/opt/gcc322) breaks lots of already existing standrd applications. Removing /opt/gcc322 (rm -fr /opt/gcc322) completely restores all the broken functionality. OK. First try: install gcc and do ldd on some broken app. Does it use this gcc's lib? Please post the output. Also 'env | grep LD', 'cat /etc/ld.so.conf'. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: kernel patches
On Fri, May 27, 2005 at 01:43:41AM +0300, Diego Iastrubni wrote: Hi, I decided that it's time to update my kernel. Since I like working hard, I usually compile my own kernels. What I usually do is use the scripts/patch-kernel command found in linux2.4 (at least...) Now, recently Linus decided to move into a 4 numbers versions, and it seems I am having problems with it. I checked several times, and I always find that 2.6.12.2 fails to patch. (upgraded from 2.6.0 into 2.6.11 and then run the You meant 2.6.11.2. script again). I suppose I am missing something. Can anyone enlighten me? [EMAIL PROTECTED]:/usr/src/kernel/linux-2.6.0$ ./scripts/patch-kernel . ../ Current kernel version is 2.6.0 Applying patch-2.6.1 (bzip2)... done. Applying patch-2.6.2 (bzip2)... done. Applying patch-2.6.3 (bzip2)... done. Applying patch-2.6.4 (bzip2)... done. Applying patch-2.6.5 (bzip2)... done. Applying patch-2.6.6 (bzip2)... done. Applying patch-2.6.7 (bzip2)... done. Applying patch-2.6.8 (bzip2)... done. Applying patch-2.6.9 (bzip2)... done. Applying patch-2.6.10 (bzip2)... done. Applying patch-2.6.11 (bzip2)... done. [EMAIL PROTECTED]:/usr/src/kernel/linux-2.6.0$ ./scripts/patch-kernel . ../ Current kernel version is 2.6.11 (Woozy Numbat) Applying patch-2.6.11.1 (bzip2)... done. Applying patch-2.6.11.2 (bzip2)... 1 out of 1 hunk FAILED -- saving rejects to file Makefile.rej Where did you take patch-2.6.11.2 from? IIRC, they (2.6.11.x) aren't incremental, but are all based on 2.6.11. There is a subdir incr with incremental patches, or simply take only the last one - that is, .., 2.6.9, 2.6.10, 2.6.11, 2.6.11.10 . -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Acting against anti-file-swapping Lawsuits in Israel
I do not believe I enter this pointless thread. Oh well. On Tue, May 24, 2005 at 11:11:01PM +0300, Geoffrey S. Mendelson wrote: On Tue, May 24, 2005 at 09:32:03PM +0300, Tzafrir Cohen wrote: You seems to have missed Shlomi's valid point here: Infriging copyrights is not exactly stealing, because the copyrights holder is still left with the goods. It may be illegal, immoral and/or publically non-hygienic, but it is still not stealing just as much that rape is not murder. It's taking of something that you don't have permission to. If that is not stealing, what is? It's not taking, it's copying. If you have an apple and I take it, you do not have it anymore. But if you have a CD and I copy it, you still have it. I did not take it. There is no need to start a new discussion about the definition of taking - if you do not agree that it's not stealing, you'd also claim it _is_ taking. The term intelectual property is intended to bind together a set of quite unrelated legal terms and make them appear similar. Patents are originally intended to promote the generation of knowledge by revealing new technologies (not making them trade secrets). Can you use a patent description today to actually replicate the invention? Does a typical patent has any use after its protection period has passed? Yes. It documents an invention. Once the period of the patent has passed, the invention passes into the PUBLIC DOMAIN. At that time you are free to copy it, manufacture it, sell it, whatever you want. And if someone sues you for doing it you can show the judge the patent and that ends their case. Tzafrir's question wasn't about the law, it was practical. Many patents today are too vague and general to be of any use. I do not have good concrete examples, sorry. Copyrights had a similar goal, but that goal has long been eroded. And worse, they originalll only applied to distribution, but now they seem to also apply to the use (with the lousy but legal excuse that every practical usage with a computer involves copying). It's not a lousy excuse. It boils down to the fact that you are buying a LICENSE to use the software, not the software. It's a two edged sword, It wasn't the point. It was about whether you can simply use it. E.g. there are attempts to create time (or similarly) limited ebooks and movies. Do you seriously agree that this was the original intent? Or that it should be like a book (that is, once you are legally an owner, you can read it as many times and for as long as you want)? you can sue someone for stealing music or software and you can sue someone for violating the GPL. Note that without copyright laws the GPL would not exist. It would be a request, but your claim would have no teeth. GPL violations are copyright violation suits. That's clear. I do not know what RMS's opinions about the GPL are, but I am pretty sure many FOSS developers using GPL would rather there was no Copyright law and no GPL rather than both. For a specific example, look at the bugroff license. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Lilo boots slow...
On Sun, May 22, 2005 at 11:01:40AM +0300, Ira Abramov wrote: Quoting Alex Alexander, from the post of Sun, 22 May: Grub worked fine for me on a RAID-1 dual SATA machine. Debian Installer (RC3 - pre-final) installed grub on the first drive and the machine boots like a charm. You just have to grub root (hd1,0) setup (hd1) to make the second harddisk bootable (in case the first one fails). just have to is not good enough. not documented enough, and what's with having to learn yet another commandline with non-standard disk enumeration and partition numbering? this MAY be excellent for someone who uses it daily maybe but I need it only once every few weeks or months, Muggles will only use it once a year or less. Why should someone start searching non-existant info pages each time?! the lilo.conf file is so much simpler that I rarely needed the manpage other than for smart menucoloring or password features, which are definitely perks more than basic features. In grub even the install procedure is crual and unusual, not to mention the odd configuration that may or may not follow :) Noone forces you to use grub. Use lilo. Just a small note - grub2 is in active development. That's probably the main reason that grub isn't very well maintained. You might want to try grub2 after it's mature. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: hebrew in vim's commands+unprintable chars
On Sat, May 21, 2005 at 09:07:05PM +0300, avraham wrote: Hi Tzafrir, 1-It's the second time that in your answer to meyou recommend to use UTF-8. It's really high time for me to understand ?? , at the very least. Can you send some pointer(s) ? Google for 'unicode unix'. First hit is the 'UTF-8 and Unicode FAQ'. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: How do you remotley manage a Win2k3 standalone box from linux ?
On Fri, May 20, 2005 at 05:36:24PM +0300, Maxim Vexler wrote: Hello list, I have a windows server that operates as standalone web server mysql db. The only remote access that currently possible to the server is using ms-RDP protocol (Remote Desktop). I think that this is insecure (I am not aware of any recent RDP exploits but just in case...). So I guess the question is 2 parted : 1. How to open RDP connection from linux box (gentoo) to the server. rdesktop 2. What would you recommend installing on the server for secure remote administration? If there is a trusted linux machine on the same (trusted) subnet, simply use ssh tunneling. You might manage to do this even with an ssh server on Windows (and so ssh directly to it and use it for both minimal remote administration and tunneling), but I did not try that. I am interested, though - if you do please tell me (us? I don't know how other list members consider this). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: mandrake su curiousity
On Fri, May 20, 2005 at 02:57:40AM +0300, Amit Aronovitch wrote: Doing 'strings /bin/su' confirmed my suspicion - the string XAUTHORITY does appear there. It's probably done by pam. Look around /etc/pam.d. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: adding text to beginning of a file
On Wed, May 18, 2005 at 03:35:00PM +0300, Micha Feigin wrote: On Wed, 18 May 2005, Micha Feigin wrote: On Tue, 17 May 2005, Aaron wrote: Hi all, I was able to add text to the end of all the files in a directory with: do cat end $foo done but I can't figure out how to add the contents of the file beginning to the beginning of all files in a directory. One way to do it would be for file in * ; do cat $file end $file done of course I got that the wrong way around and would go into a loop if it were the right way (or error). Not exactly. You wouldn't go into a loop, and if cat is naive, also not get an error. You'd simply get in $file only the contents of 'end'. That's because ' $file' is interpreted by the shell, even before running cat. The shell opens the file and truncates it. cat would read it, it will be empty, and then go on and read 'end' and cat it into file. However, I was not sure cat is naive. So I tried this. On Linux (2.6, debian sid), Solaris 8 and Irix 6.5. On all three of them, cat noticed and said something. Linux (actually, GNU cat): % cat f1 end f1 cat: f1: input file is output file Solaris: % cat f1 end f1 cat: input/output files 'f1' identical Irix: % cat f1 end f1 Input/output files 'f1' identical I was curious to know what exactly cat did. It turns out that it skipped f1. To know this (without looking at the sources), I simply did ' f1' instead of ' f1'. If it did read it, it would double it (catting itself to the end of it). But it didn't - I simply got f1 and then end. what you want is: for file in * ; do cat begining $file $file.tmp mv $file.tmp $file done Since I already took the time and rudeness to reply, I'll add: Your solution is mostly correct. As others said, all of these things aren't safe. If you get used to doing things both unsafely and without awareness, you'll end up loosing data. Double check whatever you do. I already wrote my preffered solution (of course there are many good ones). It's left as a simple excersize to the reader why I prefer mine over yours. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: adding text to beginning of a file
On Tue, May 17, 2005 at 09:26:40PM +0300, Aaron wrote: Hi all, I was able to add text to the end of all the files in a directory with: do cat end $foo done but I can't figure out how to add the contents of the file beginning to the beginning of all files in a directory. I know this is something simple but I can't remember how to do this. The reason you do not remember is that it's not simple. Actually, it's not possible. Unix doesn't have an API for inserting data into a file, and therefore there is no small utility (or shell syntax) that does that. There are few ways to solve this. One is to use some workaround. E.g. do mv $foo $foo.orig cp beginning $foo cat $foo.orig $foo done Then examine the new files, make sure they are ok, and rm *.orig A better way is to do this semi-automatically. E.g. with a preprocessor, every time you run the program that converts this file to midi. I would write a small Makefile that converts the files to midi and do this in the Makefile. The best way is to use a language that has tools for this, e.g. #include in C. I do not recognize the one you currently use - it might have this possibility. I can recommend a language called mma. I played with it a little and it's quite nice. - This is end - } \paper { } \midi { \tempo 4 = 60 } } this is beginning \version 2.2.0 \header{ title = Nigun Hisvaadus composer = meter = } #(set-global-staff-size 18) \score { \notes { \key e \minor \clef violin --- --- Thanks Aaron -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Is open source more secure? [was Re: Moving to Linux]
On Sat, May 14, 2005 at 10:33:42AM +1000, Amos Shapira wrote: On 5/13/05, Yedidyah Bar-David [EMAIL PROTECTED] wrote: common as a server. So the crackers develop means to break linux servers. If/When linux is very common on the desktop, you'll start seeing the same there. Same flawed FUD used by the MS camp. Apache is the most common web server in the world and still it is cracked much less than IIS. Same situation (server, public net) still different results. To get to even more philosophical ideas - I must say I do not accuse MS for behaving the way it behaves. I accuse the users. Users want Bread I *DO* accuse MS. They kept reasoning that people want convenience and that they must compromise security to give it to them. I never intended to imply that in order to give convenience you have to compromise security. What I wanted to say is that MS gives people what they want. Most people do not care much about security, so MS does not invest in it. Of course doing both is possible, as you showed. Or maybe I should say, people do not care _enough_ about security. They (most of them) would not consider moving off Windows only because it's not (less?) secure. I also guess that most people say to themselves that Windows can't be _that_ bad, or else less and less people would use it. A billion flies can't be wrong. Eat shit!. All of this, BTW, refers to a few years ago. Recent years changed both the understanding of the average users and their options. But I am not sure that we are close to World Domination (tm). I am personally waiting for this since 1997, when Byte first had the item Linux in a Gray Flannel Suit. Oh was I excited! -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Is open source more secure? [was Re: Moving to Linux]
On Sat, May 14, 2005 at 09:10:59PM +1000, Amos Shapira wrote: On 5/14/05, Yedidyah Bar-David [EMAIL PROTECTED] wrote: On Sat, May 14, 2005 at 10:33:42AM +1000, Amos Shapira wrote: On 5/13/05, Yedidyah Bar-David [EMAIL PROTECTED] wrote: common as a server. So the crackers develop means to break linux servers. If/When linux is very common on the desktop, you'll start seeing the same there. Same flawed FUD used by the MS camp. Apache is the most common web server in the world and still it is cracked much less than IIS. Same situation (server, public net) still different results. To get to even more philosophical ideas - I must say I do not accuse MS for behaving the way it behaves. I accuse the users. Users want Bread I *DO* accuse MS. They kept reasoning that people want convenience and that they must compromise security to give it to them. I never intended to imply that in order to give convenience you have to compromise security. What I wanted to say is that MS gives people I was stressing the point that it is MS's fault (as opposed to your I do not accuseMS quoted above). They made the decisions to provide insecure software. What could their users do about MS's software? what they want. Most people do not care much about security, so MS does not invest in it. Of course doing both is possible, as you showed. So what's wrong with what users did? They eat what MS feeds them. Microsoft is a company. It makes money. That's its goal. There are good companies, there are bad ones. This one is bad. So what? Noone forces people to be customers. Of course MS puts a very strong pressure on people to be. So strong most of them do not even realize they have a choice. I do not claim I am happy about the situation. But I do not think MS could have arrived to where it is today if most of the people in the world would behave and think like we (without defining what exactly is we) do. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Is open source more secure? [was Re: Moving to Linux]
On Fri, May 13, 2005 at 10:18:49AM +0300, Ori Idan wrote: I think this is an academic debate if GNU/Linux is more secured or not. For the simple people, let us look at the facts: 1. When was the last time any of this list members has seen a virus in his GNU/Linux desktop? (I guess the answer is never) 2. When was the last time you had a spyware in your desktop? (again the answer is never) So the end result is: GNU/LINUX IS MORE SECURED. No, the end result is that the members of this list are more security-aware than the average. Much more so. I'd make a wild guess - even if all of the members here used XP/MS Office regularly, they would still have much much less virii/spywares than the average. OTOH, if 80% of the world's computer users used Linux, you'd see more viruses and spyware in Linux. Maybe not as much, maybe with less average damage, but while I agree to your conclusion, I do not agree to the reasoning. This reminds me of Paul Graham's articles, in which he claims that LISP programmers are better. But why is it so (whether or not you agree to the conclusion)? There are at least two opposite reasons: 1. Because programmers that learned LISP become better 2. Because good programmers prefer LISP when they come to know it. Just to prove my point - everyone here will agree that putting a default install of most major distros open in the net without some kind of firewall or hardening will very quickly make it broken into (I know about exceptions, no need to remind me). Why? Because linux is very common as a server. So the crackers develop means to break linux servers. If/When linux is very common on the desktop, you'll start seeing the same there. To get to even more philosophical ideas - I must say I do not accuse MS for behaving the way it behaves. I accuse the users. Users want Bread and fun (LECHEM VESHA`ASHU`IM). Users do not want freedom. They do not want security. Nor power, nor robustness. They want a lot of software, that doesn't require them to read anything or to think in order to use it (note I did not use the shorter term easy to use, I know it won't pass some tests). So that's what MS sells to them. It's not that I agree to all of MS's policies - but I think this is the root cause. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Moving to Linux
On Tue, May 10, 2005 at 02:06:30PM +0300, Shlomi Fish wrote: Heh. It is also important to understand that hardware played part in shaping the evolution of editors. When UNIX started, computers wrote output to line printers on paper, (very slowly). So people created editors like ed, where you typed a line and executed it. Then came terminals without relocated cursor (IIRC) and so ex evolved. Once terminals where the cursor was relocatable evolved, Bill Joy created vi to be used. It should also be noted that the keywords when vi was created did not have many modifier keys that we now take for granted. They didn't have the Alts for sure, or the F-keys and they may not have had the Ctrl modifiers either. They did not have the Num pad much less the IBM PS/2 cursor keys and concentrated Home/End/Insert etc. scheme. I do not know what exact keyboard Joy used while writing vi, but my guess is that it wasn't lacking modifier keys. Rather, at the time, modifier and control keys were non-standard. There were many different keyboards with many different keys. So he decided to use the lowest common denominator. An antithesis is emacs, which some claim is an acronym for 'Escape Meta Alt Control Shift'. For a nice description of the subject, read the entry 'space-cadet keyboard' in the jargon file. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Moving to Linux
On Mon, May 09, 2005 at 08:16:40PM +0300, Shlomi Fish wrote: On Monday 09 May 2005 15:23, Amit Aronovitch wrote: If your'e Micorosft, you might create a central distribution source carrying Windows, Office, several games and tools, but what about Photoshop? Doom3? Acrobat Reader? WinZip? You can't legally distribute those without special contract with the authors (well, you can always buy some companies, and put others out of business ;-) ). Of course, you could add some Free Software in your distribution too - but you can't add GPL-licensed stuff (and GPL is the most common OSS license). If you do add GPL stuff, you'll have to make all the other stuff open source too - so the commercial parts are out - you can't supply Office Windows. That's bullshit. Microsoft (for example) can distribute updates to GPLed software along with their own proprietary software without any restriction whatsoever. As long as the GPLed components install to different files, there's no restrictions whatsoever on the distribution medium of GPLed software. On my hard disk I have Opera[1] which is proprietary along with gcc which is GPLed. If I make a tarball out of both, would it make Opera GPLed? Or am I breaking the law? Of course not. Do you want to say that Debian is breaking the law by supplying updates to GPLed program from the same medium as Open Source Software (which may not necessarily be Free according to the FSD), under a non-compatible license? Hell no. The only restriction Microsoft have is that they supply the sources to the GPLed program on demand or on their web-site. (or at least point someone to where they can find them IANAL). That and if they make modifications to their sources, they must distribute these modifications or the modified sources. Whether Microsoft indeed want to supply updates to GPLed and other software that is free as in speech from its updates source is a different question. But if they do decide to, it will be fully legal. Actually, MS did distribute GPLed software. NT Resource kit contained perl. IIRC with sources. I don't know if recent RKits continue this tradition. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Moving to Linux
On Mon, May 09, 2005 at 11:30:06PM +0300, Oron Peled wrote: On Monday 09 May 2005 21:34, Yedidyah Bar-David wrote: Actually, MS did distribute GPLed software. NT Resource kit contained perl. IIRC with sources. I don't know if recent RKits continue this tradition. 1. FALSE: perl license is not GPL (it's under the Artistic License) It's actually dual-licensed, either GPL or artistic, at your option, as far as I know. I cannot say I fully understand either, but choosing the artistic license would still require providing sources or changing the executable name (which they didn't do), at least as far as I understand. 2. TRUE: MS does distribute GPL software (the Services For Unix) with the sources as required. That's a bit weird - why didn't they take BSD sources? Let's all stick to facts please. Yes, I agree. The point I wanted to make is that this isn't hypothetical. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: undeterministic zip?
Hi all, This isn't more on topic than the original subject, but I allow myself to share this very valuable information. From the NEWS file of GNU tar version 1.15 (which isn't in Debian unstable yet, I do not even remember how come I ran into this): version 1.15 - Sergey Poznyakoff, 2004-12-20 * Compressed archives are recognised automatically, it is no longer necessary to specify -Z, -z, or -j options to read them. Thus, you can now run `tar tf archive.tar.gz'. :-) It's irrelevant for Ira's question, as tar can't guess if you want to compress an archive. But now it knows to decompress it. If it had this option a few years ago, it would sure have saved me a few tens of times re-running tar commands with j and z replaced. But of course I prefferd to suffer than to think a bit and write such a patch myself (or even a wrapper script). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Any problem with hamakor.org.il?
Hi all, I can't access hamakor, from two different networks. Anyone knows why? Sorry if a planned downtime was already announced and I missed it. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: VMware GSX host file systems
On Thu, Apr 28, 2005 at 10:52:20PM +0300, Gil Freund wrote: On 4/28/05, Marc A. Volovic [EMAIL PROTECTED] wrote: XFS is optimized for sequential access performance. Just to clarify, do you mean batch type processing? Without knowing, I guess Marc meant audio/video, which is a large part of SGI's users. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: unstable vs. sarge (was: Re: culmus 0.10 on Debian Sarge?)
On Tue, Apr 26, 2005 at 03:28:31PM +1000, Amos Shapira wrote: On 4/26/05, Tzafrir Cohen [EMAIL PROTECTED] wrote: Baruch's answer on why not installing it) and sid should remain unstable anyway. What do you mean by the last sentence? Won't sid be promoted to testing when sarge is released? If not then what is the defining milestone to trigger that? sid is the name of unstable. It will always be. The next release after sarge will be called etch. This means that when sarge is released, stable will be sarge, testing will be etch, and unstable will remain sid. Interestingly, I was not sure about the name etch. So I googled for sarge woody potato and the first non-mailing-list result was not a .debian.org page but the wikipedia page for Debian. It might be a good page to read if you do not know Deian well. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: [OT] FOSS OCR solutions?
On Thu, Apr 21, 2005 at 01:15:41AM +0300, Amit Aronovitch wrote: Thanks alot for the reply. I'll try having a look at both tools. Yedidyah Bar-David wrote: Description: --- We have a large database of scans consisting of numerical data, which need to be accurately converted to text, with minimal human intervention. The scans are low quality (~200dpi, some with dirt (random dots), some misaligned, some upside-down etc.). Are they fixed-font? Yes. At least the pages I saw (data came from several sources). So it'll be easier. hebocr and clara do only fixed-font. However, the data is numbers only, and comes in ordered tables with a fixed number of columns (no border/cell lines at all). Numbers are of fixed width. Most of the cells are full + there are certain rules which can be applied (i.e.: if a row (entry) has a number in column A, it must contain data in column B too). I think no OCR has a language to say these rules, but some are script- friendly enough to allow this easily. hebocr has a log file in which it reports almost every important thing (including the position in pixels of every char) so you can use it instead of the raw text output. I was thinking of writing some script/program that calls some lower level API (e.g. call some algorithm for finding cell positions, then use above mentioned rules to improve this suggestion, then once you know where the cells are, and you know each one should contain a number with a known number of digits - call the actual digit recognition function for each cell). Do you think it would be simple enough to write such an algorithm with hebocr? (e.g.: does it have seperately callable region-finding and char-recognition functions?). It does have specific sets of functions (it's C - no classes, packages, etc.) for dealing with different tasks. It has a short text file describing how it works internally (written by me) and, as I said, someone wrote a much longer tutorial, which I don't know if is available online, I'll ask his permission. The data must be placed in the right columns: we prefer having a known number of unknown digits (even a few incorrect ones) than 'missing' or 'added' numbers. Preferably, we should be be able to get a certainty measure for each digit. I think every OCR can give you that. I know mine does. Well, I saw some commercial tools that just don't have this (probably they have it in some lower level but don't export it to user. With closed source - this is a dead end...). The solution should have some adjustable configuration parameters, such that we'll be able to reprocess selected 'important' pages (based on initial results) with some human adjustmants. What parameters do you think about? clara has lots of such params, but I find it a bit uncomfortable to use, because you need the UI to play with them etc. Look at it, though. hebocr has very few such params. The main way to affect it is the font you provide (or build, using its output). Of course there are many things you can also do outside the OCR - e.g. if the scans are gray, you can change the threshold of white/black and have a dramatic change in the accuracy. Few OCRs also read gray directly, but it's not common. Theyre b/w. I was thinking of stuff like thresholds for some filters and preprocessing algorithms. e.g. despeckle parameters (if page was noisy you could change), thresholds for cell/area detection algorithms (if for some reason you see that the algorithm missed some cells - you can tweak it). I don't think FOSS OCRs deal with such things. There are other tools for this, and you can glue them together with a small script. I know I personally scanned in grayscale and only played with the threshold for white/black. In my case there were pages with different blackness, and so eventually I wrote a little script that computes the threshold so that they all have more-or-less the same coverage (17% in my case). Yes - probably would be uncomfortable to play with (UI could help), but you'd only do this for 'important' pages. it should still be faster than typing these pages manually. How large is the database? If it will take only a few hours (maybe even a day or two) to manually type it, I think no free OCR will do better. To beat that you need one that needs no training and has very good results. These two needs are hard to provide together. And numbers are very fast to type, with very few mistakes, compared to text. It's a specific niche in the job market - I know because I know people who worked in this, wanted to move to OCR, and didn't because it wasn't good enough. As a very rough guess - 1000 pages (and their'e pretty dense with digits) - would simply require too many typists for too much time. We'd probably need some initial scan anyway - to prioritize the pages
Re: [OT] FOSS OCR solutions?
On Wed, Apr 20, 2005 at 01:07:28AM +0300, Amit Aronovitch wrote: Hi list, I'm looking for a free software tool for solving a specific OCR task (details below). I've googled up several projects, but because of time limits I'll not be able to give them a fair try. If you have some experience with such tools, I'll be grateful for any tips. Specificly - how hard it would be to use them for solving the specific task at hand. Note that since the task is not a standard one, I don't expect any tool to give good enough results out-of-the-box. Instead, I expect it to be modular, configurable, and scriptable enough for writing a task-customised solution in short time. I naturally know well only the one I wrote myself, hebocr. I am not sure if it's good for your needs, but I _am_ sure it's the smallest one in lines of code. So if you'll need to change it, it will probably be the easiest one to try. I played a bit with clara and it seems quite good. The latest news about hebocr are that someone (not me) invested some time in it and now it again works (even on cygwin) and has a tutorial. It's still considered dead by me, unless someone wants to take over. Description: --- We have a large database of scans consisting of numerical data, which need to be accurately converted to text, with minimal human intervention. The scans are low quality (~200dpi, some with dirt (random dots), some misaligned, some upside-down etc.). Are they fixed-font? However, the data is numbers only, and comes in ordered tables with a fixed number of columns (no border/cell lines at all). Numbers are of fixed width. Most of the cells are full + there are certain rules which can be applied (i.e.: if a row (entry) has a number in column A, it must contain data in column B too). I think no OCR has a language to say these rules, but some are script- friendly enough to allow this easily. hebocr has a log file in which it reports almost every important thing (including the position in pixels of every char) so you can use it instead of the raw text output. The data must be placed in the right columns: we prefer having a known number of unknown digits (even a few incorrect ones) than 'missing' or 'added' numbers. Preferably, we should be be able to get a certainty measure for each digit. I think every OCR can give you that. I know mine does. The solution should have some adjustable configuration parameters, such that we'll be able to reprocess selected 'important' pages (based on initial results) with some human adjustmants. What parameters do you think about? clara has lots of such params, but I find it a bit uncomfortable to use, because you need the UI to play with them etc. Look at it, though. hebocr has very few such params. The main way to affect it is the font you provide (or build, using its output). Of course there are many things you can also do outside the OCR - e.g. if the scans are gray, you can change the threshold of white/black and have a dramatic change in the accuracy. Few OCRs also read gray directly, but it's not common. How large is the database? If it will take only a few hours (maybe even a day or two) to manually type it, I think no free OCR will do better. To beat that you need one that needs no training and has very good results. These two needs are hard to provide together. And numbers are very fast to type, with very few mistakes, compared to text. It's a specific niche in the job market - I know because I know people who worked in this, wanted to move to OCR, and didn't because it wasn't good enough. Sorry for the somewhat OT issue, but I know of no other forum I expect to answer this in short time. (besides, the preferred target platform IS linux :-) ). There was a long thread about OCR in whatsup around a year ago. You can also ask there. Amit A. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: resizing /dev/shm
On Wed, Apr 13, 2005 at 03:35:03PM +0200, Hetz Ben Hamo wrote: Hi people, In most cases, when I'm doing df -h, I find that /dev/shm (shared memory) can use up to about half of the RAM my machine has (in my case - 506MB). Does anyone knows a way to increase this size? I tried few tricks with /etc/fstab and it doesn't seem to be helpful.. Any suggestions - are more then welcome. In Debian you can set this through /etc/default/tmpfs. Don't know others. To set it manually, do something like mount -o remount,size=70k /dev/shm You can probably easily find where your distro sets this and change it permanently. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: strange crush
On Mon, Apr 11, 2005 at 10:04:04AM +0200, Erez Doron wrote: hi i have a mythtv/fc3 PVR, or actually, i had one :-( yesterday, after i went against the 'if it works, dont fix it' and did a 'yum update', my pvr rendered usless. i rebooted, and got: 'unexpected inconsistency, please run fsck' on my / filesystem (ext3) so i did 'fsck /dev/sda3' ( it is a sata disk), and answered a lot of 'y'. then i rebboted again, and stil get the 'unexpected inconsistency' etc. it doesn't mattar how many fsck i run, i get to same thing again and again. Maybe some RAM/MB/CPU problem? I'd run memtest86 just to make sure. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Failure in running system from within perl within apache
On Thu, Apr 07, 2005 at 05:09:00PM +0300, Shachar Shemesh wrote: Hi all, I'm building a small web application. As I want this app to make changes in the system, I'm using a suid (non-root) perl executable to carry out most of the actual operations, and the application (read - apache) runs this executable. So far, so good. One of the operations I need carried out is creating an SSH key. I use the following syntax inside the perl script: if( system(ssh-keygen, -q, -b, $1, -t, dsa, -f, /home/user/.ssh/id_dsa, -N, , -C, $2 )==0 ) { If I run the perl script directly, everything is great. If I try to run it from the web server, however, the system isn't carried out. I thought it may have to do with system trying to rely on a shell which the apache user doesn't have, but the man page for perlfunc says that a shell is only required if the other syntax for system is used, and that execvp is used for this syntax. I also tried switching to the other syntax and redirecting the output to a file. Nothing. The file is not even created. I tried replacing ssh-keygen with /usr/bin/ssh-keygen, but that did not solve the problem either. In short, I'm fairly running out of ideas as to how to debug this. I can't seem to find a way to test what is going wrong. Ideas would be most welcome Did you try strace and see that it really doesn't try to exec? If you do, note you can't strace a suid exec. To do this, strace -p as root. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Current directory in PATH [Was: A second glibc on Linux]
On Thu, Mar 31, 2005 at 04:01:50PM +, Oleg Goldshmidt wrote: Well, many years ago (not 20, but still quite a few) I came to the conclusion that as a regular user I want . in my PATH, but only in the last position. It is a matter of convenience, and the security problem associated with it is limited as long as I am wearing my unpriviledged user hat. I never add . to my root user PATH, because I am paranoid. This is the compromise I found suitable for me, and it has worked very well for me for all these years. I am not a newbie, of course, and it does not take much effort not to call executables test or use tar as an abbreviation for targil ;-). Of course, the system tar will be invoked, and most likely fail since the command-line arguments will match its expectations only by a rather improbable coincidence. Fine for you, but what would you recommend for a new user (a learning programmer, not a naive user)? BTW, today was my last day as a sysadmin at tau. So gradually I will stop thinking about these matters. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Current directory in PATH [Was: A second glibc on Linux]
On Thu, Mar 31, 2005 at 11:17:22PM +, Oleg Goldshmidt wrote: [EMAIL PROTECTED] (Yedidyah Bar-David) writes: Fine for you, but what would you recommend for a new user (a learning programmer, not a naive user)? In my mind, this is a good scheme. Not the only one possible, but I see advantages in using it. Putting . in PATH is a security issue that a learning programmer should understand. The difference between an unpriviledged user and root should also be understood. Avoiding shadowing of system utilities is a general principle that a learning programmer should also think about. Well, I am not sure. I use unix for 13 years now, almost never had . in my path, and am very comfortable with this. BTW, today was my last day as a sysadmin at tau. Good luck in whatever you do next. Thanks. So gradually I will stop thinking about these matters. Are you leaving computing altogether? I mean, I have never been a sysadmin at TAU but I think about these issues... ;-) I do not leave computing, but I am not sure I will soon (e.g. in the next few years) work again in a place serving studying programmers. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: high load ?
On Wed, Mar 30, 2005 at 09:25:29AM +0200, Shachar Shemesh wrote: guy keren wrote: [snip] Processes should never spend too much time in the D state. The very fact that certain activities mean you are almost guaranteed to see processes in the D state means there are bugs in the kernel. However, these are unrelated to system load. A process in the D state is not consuming CPU, and will not prevent my shell from getting to the top of the ready queue and getting the CPU. D state processes distort the load average number. Agreed. I know there are people that patch their kernels to not count D state processes in the load average. [snip] You may have to tweak the numbers a bit, but it seems about right. A different question is whether, under this scenario, the load average is still the right metric to look at? I think it is. If the load average is 2, my shell still have quite a queue to wait for being actively processed, and the responsiveness will depend on the time slices alloted to each process. This is so only if you have load only on the CPU. If, for example, you only have one process running, but which does a lot of paging, your load average will be =1, but the responsiveness will be quite bad, as your shell will probably be paged out when you press a key. OTOH, today's CPU are very fast, and in most cases, if you run a few 'for (;;);' in the background, an interactive shell user won't notice, while the load average will equal the number of such loops. Much more annoying for our own users is amd mounts that sometimes take a lot of time or stuck. I know this isn't kernel-related, but it does hurt a lot responsiveness. I personally do not know any single number that can tell on unix what the expected responsiveness is. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Floppy image loaded from grub?
On Wed, Mar 30, 2005 at 11:20:49AM +0200, Kfir Lavi wrote: On Wednesday 30 March 2005 02:35, guy keren wrote: On Tue, 29 Mar 2005, Kfir Lavi wrote: Hi, i would like to install netBSD. They have floppy image for this installation. I don't have floppy or cdrom attached to my computer. Can i load the image from the hard disk with grub? other program? yes - look for info on loading 'memtest' from grub - i think some distributions have this configured by default, and i _think_ this is done by loading a floppy image. Yep, i'm using memtest, and in debian its automaticly, but... The size of floppy image considered as big because grub and lilo thinks its a kernel, so it will not boot. memtest seems like a linux kernel. That is, it's self-bootable (just as linux used to be not too long ago) and is also bootable by linux bootloaders (lilo, grub, etc.). It won't help you to load a general floppy image. As I said before, try memdisk. It's a small program that looks like a linux kernel and emulates a floppy for its payload (which IIRC is loaded as an initrd). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: high load ?
On Wed, Mar 30, 2005 at 10:36:07AM +0200, Shachar Shemesh wrote: Yedidyah Bar-David wrote: You may have to tweak the numbers a bit, but it seems about right. A different question is whether, under this scenario, the load average is still the right metric to look at? I think it is. If the load average is 2, my shell still have quite a queue to wait for being actively processed, and the responsiveness will depend on the time slices alloted to each process. This is so only if you have load only on the CPU. I'm sorry, I don't see how the number of CPUs enter into it. I'm not sure whether the load average should be divided by the number of CPU's or not in order to get a consistent number, but that's besides the point. I was merely trying to show a situation in which the CPU is relatively idle, but the load average is high, and then ask whether we are still more interested in the load or the idle time. I never ever talked in this thread about number of CPUs. If, for example, you only have one process running, but which does a lot of paging, your load average will be =1, but the responsiveness will be quite bad, as your shell will probably be paged out when you press a key. Not if I repeatedly hit keys. That said, obviously both load average and Even if you repeatedly hit keys, you might have bad responsiveness due to thrashing. The VM subsystem did get better in 2.2-2.4-2.6, but still can't be perfect. CPU state are CPU metrics. Total system performance is affected by other factors, such as disk speed, memory usage, network latency etc. These are not reflected in the load average at all. If a process is waiting for disk or network it won't be in the Ready queue. I'm not sure about swap though. As for CPU stats, it may be somewhat reflected in the amount of time the CPU spent in system vs. user, but not in the idle count. OTOH, today's CPU are very fast, and in most cases, if you run a few 'for (;;);' in the background, an interactive shell user won't notice, while the load average will equal the number of such loops. What does CPU speed have to do with it? A faster CPU will perform many more no-op loops than a slow one, but would still take 100% CPU. The scheduler today may be smarter and give my shell a higher priority due to its interactiveness, but that, again, has nothing to do with the CPU's speed. It seems to me that the only thing related to speed affecting the perceived responsiveness of my shell is going to be the time slice alloted to each running process. IIRC you can change the time slice independent of CPU speed or architecture. It's true that some architectures define it to be smaller by default. But I don't think that's the point. The main thing you get with a fast CPU is that context switches are faster. Much more annoying for our own users is amd mounts that sometimes take a lot of time or stuck. I know this isn't kernel-related, but it does hurt a lot responsiveness. A. See above for non-CPU related problems. B. I know that the NFS code used to have TONS of locks held for far far far too long. Any network latency spewed processes in D state by the bucket load. That is very much kernel related. I don't know how things are today. I did not check thoroughly, but I do have a feeling things are getting better over time. But other things changed here besides kernel versions. I personally do not know any single number that can tell on unix what the expected responsiveness is. RTFM time(1) time doesn't measure responsiveness, but throughput. IIRC people did write benchmarks that measure responsiveness, and I agree that they give you, in the bottom line, a single number, but they are (probably) heavy to run. They are benchmarks. What I meant was to say that among the things that unix counts all the time (and therefore are cheap to count), no single one is enough to tell how responsive the system is. load average is the common number for this task, and is in many times good enough, but as I said, not always. In general you do have to look at vmstat, cpu state(s), etc. to get a picture of what's going on. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: A second glibc on Linux ( there's a keren in the darkness )
On Wed, Mar 30, 2005 at 05:40:41PM +0200, Nadav Har'El wrote: On Wed, Mar 30, 2005, guy keren wrote about Re: A second glibc on Linux ( there's a keren in the darkness ): On Wed, 30 Mar 2005, Oron Peled wrote: To summarize: the folk tale about avoiding commands named test (or Nee, for that matter) is like trying to cure a virus with Aspirin. this is wrong, as it does not take into account the fact that a newcomer is sometimes accustomed to the DOS way, where '.' is always in the PATH, implicitly. as such, they don't think they need to use './test'. with other programs, they would get a 'command not found' error. with a program like 'test', they will get such an obscure error message that they'll have no idea how to even begin debugging it. Or worse: they don't get any error message at all. test with zero or one parameters just does nothing, and returns. I've seen this happen to at least 3 unsuspecting newbies... P.S. I disagree that having the current directory in the path is only the DOS way. It has always been the Unix way too, and I still like it to this day. I think that many books even recommended (or perhaps even still recommend) that non-root users have a colon starting their path, meaning that the programs in the current directory take preference. This is useful for programmers, but indeed is probably less useful for non-programmers, which is why this practice fell out of favor over the years. Nadav, of all the things you said until today, this is surely the one I least agree with you about. Can you show a concrete example of such a book? I personally prefer to explain again and again to new users here (most of which are programming students) why putting . in the path is bad and that they should ./ rather than tell them how to add it (or even, as you suggest, to make it the default). Just to give a small proof that I am not entirely wrong, in tcsh there are two relevant compile-time options - one is to move '.' to the end of the path and one is to omit it altogether. So enough people found it useful to add to tcsh. I am also not sure why it's so much more useful for programmers. You do not have to add './', right, but you have to add a few chars for tab completion to work (because it searches the entire path). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: A second glibc on Linux ( there's a keren in the darkness )
On Thu, Mar 31, 2005 at 01:02:23AM +0200, Nadav Har'El wrote: On Wed, Mar 30, 2005, Yedidyah Bar-David wrote about Re: A second glibc on Linux ( there's a keren in the darkness ): P.S. I disagree that having the current directory in the path is only the DOS way. It has always been the Unix way too, and I still like it to this day. I think that many books even recommended (or perhaps even still recommend) that non-root users have a colon starting their path, meaning that the programs in the current directory take preference. This is useful for programmers, but indeed is probably less useful for non-programmers, which is why this practice fell out of favor over the years. Nadav, of all the things you said until today, this is surely the one I least agree with you about. Like I said, this is perhaps because I learnt Unix twenty years ago (wow, time flies...), and fashions have changed. I am still used to having almost the same PATH setting I had 20 years ago... Can you show a concrete example of such a book? Sure. The classic (but a bit long in the tooth...) The UNIX Programming Environment by Kernighan and Pike, first mentions the PATH in page 37, and gives the following example: [snip] Well, you caught me. I did not read it. So perhaps this tradition is no longer in vogue, and perhaps there are a 1001 security reasons and other valid reasons to avoid it, but it still is a generation-old Unix tradition. Not a DOS tradition :) Ok, ok ... Just to give a small proof that I am not entirely wrong, in tcsh there are two relevant compile-time options - one is to move '.' to the end of the path and one is to omit it altogether. So enough people found it useful to add to tcsh. The fact they even had to deal with this . means that having it was a common tradition, I think. This makes sense. I am also not sure why it's so much more useful for programmers. You do not have to add './', right, but you have to add a few chars for tab completion to work (because it searches the entire path). Perhaps. Although if you have a command called nu, then typing nu (2 characters) is definitely faster than ./nTAB (4 characters). Silly example? I know. Never mind. And what if you want to run /usr/bin/nu? I must admit (again showing my ignorance) that your example made me see for the first time that there is such a thing, without a manpage or anything. Part of the netatalk package, which probably isn't commonly installed, though. I'm not trying to defend the dot in my PATH. It's probably a thing of the past. I just didn't want it to be called... DOS-like. shudder :) Let's just say this: I suggest that at least on public machines, you stop having '.' in your path, and also that you do not recommend this to others. Unix (not the OS but the average user) and the Internet aren't what they used to be 20 years ago. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Floppy image loaded from grub?
On Tue, Mar 29, 2005 at 11:41:36PM +0200, Kfir Lavi wrote: Hi, i would like to install netBSD. They have floppy image for this installation. I don't have floppy or cdrom attached to my computer. Can i load the image from the hard disk with grub? other program? You can try using memdisk - part of syslinux. IIRC grub has native support for *BSD - you might try finding relevant files to load directly instead of the floppy image. -- Didi tnx Kfir = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Linux 2.6.x and RAID-1 support for a Silicon Image controller
On Mon, Mar 28, 2005 at 11:35:37PM +0200, shimi wrote: --=-iuQFSqE5IfCkytOwtoW6 Content-Type: text/plain Content-Transfer-Encoding: 7bit On Mon, 2005-03-28 at 23:11 +0200, Ilya Konstantinov wrote: shimi wrote: After looking up in Tyan's site, and looking in the specs of the MB (URL: http://www.tyan.com/products/html/tigerk8w_spec.html http://www.tyan.com/products/html/tigerk8w_spec.html%29, ), I saw that it contains Silicon Image Sil3114 SATA Raid Accelerator. I contacted the store I am planning on buying the machine from, and asked them wether it's well supported in Linux. I got a weird answer that the SATA will work, but you'll have to do the RAID-1 you're planning to do by software... Sounds like a pretty sensible answer. I would say more - it says very good things about this store. I am yet to have such an experience myself. [snip] 2. I don't need to control the Hardware RAID from within Linux, as I'll not be building/removing volumes etc on the live system... (however reading errors from the controller would be nice). I was surprised about the fact the SATA works, but RAID [completely. not only management] will not. I had the impression that when you do a RAID with hardware running it, it appears to the OS as one single device (/dev/sda, probably), and the OS doesn't even KNOW that the drive is mirrored... (that's how it was for me in the past with an Adaptec doing RAID-5 on SCSI drives, it just appeared as /dev/sda. But that was native-SCSI, and maybe SATA is different. I don't know. :)). Are you sure it's hardware raid? I am pretty sure it's not, it's software raid done by the driver. I never used Silicon Image raid, but did use their SATA. Not this chip, though - only SiI3112. And still not with 2.6/libata. Works well. Hope to get clarifications, and more importantly, success stories! :) I would personally not use its raid even if it exists and works well. The only advantage you get over native linux software raid (probably md, but other options exist) is that you keep the configuration in an OS- independant way, which allows easily dual-booting over this raid. Not useful for a server. That is, unless they added some very fancy stuff to their driver. But it's still just software. The spec you sent also says Connected to legacy 32-bit 33MHz PCI bus. I don't know what it means, but you might get better performance by connecting one disk as SATA and one as PATA (which means non-identical disks). At least I did. On some older hardware, but I would check this also today. It was on this machine (don't recall exact board model, but it's a dual xeon 2.4Ghz, 2GB RAM and this sii3112), with two 120GB disks. They were both pata, but were bought with sata convertors. So I could connect them as I wanted, and also knew they are exactly the same (in terms of same partition tables, etc.). They both gave me, when working alone, both though the sii and the ICH4, 55MB/s. When both were through sii, I only got 44MB/s. When one through sii and one through ich4, 55MB - no loss at all. When both through ich4, only 32MB/s. Which means the sii was faster than the ich4, but still not fast enough to handle both disks. If you do make some measurements, I would love to hear your results. All of this, BTW, was with zcav on the disks themselves. I did not measure md raid1 write speed (reading is only from one disk anyway). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: I must be missing something
On Wed, Mar 23, 2005 at 09:09:01PM +0200, Shaul Karl wrote: On Wed, Mar 23, 2005 at 07:46:24PM +0200, Shachar Shemesh wrote: There is no other explanation. Just about every free program on Linux keeps a configuration file in /etc. They all fairly much carry the same format. It simply defies logic that there would be no standard library to read this configuration file in. There are few such projects. I have yet to see one that has a considerable following outside its natural friends. The famous ones are probably gconf and mcop (I never used them as a programmer and know very little about them as a user, so excuse me if they are not a direct answer to the original question), and there are few others - e.g. search freshmeat for 'ini'. One more example - I have seen (but did not use myself) lua being used also for such specific uses. It's usually considered a general extension language, at the same class of e.g. tcl or guile, but is very small and has specific features for such things. Not sure I understand what you are after. Suppose there would have been # include configuration_files.h. What API would you expect to find in it? What output would you expect to get from reading fstab? timezone? whatever? I agree it's hard to define a language good enough for everything, even more so good (not mearly enough) for everything, just as there is no programming language good for everything. Assembly might be good enough for everything, and just as you would not want to use it for everything you would not want to use any specific language for conf. Does this standard library exists on other OSs? As was already mentioned, the Windows Registry. As far as I know, there are few such attempts also in various Unices. Some of them even became semi-standard. E.g. nss, which is a library for accessing the user db, etc - but I do not even know how easy it is to add a new type for your own program. IIRC there are some more extensive ones, maybe in AIX or in MaxOS X, but I do not know that for sure. BTW, IIRC, one of the visions of Be with BeOS was to clean up the unix mess, one of the aspects being to create a unified conf format. As we all know, Be didn't do well. It seems people do not really want cleanness. They want features and power. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: imitate rar adding checksum
On Thu, Mar 10, 2005 at 02:34:51PM +0200, Kfir Lavi wrote: Hi, when compressing with rar you can choose to add error correcting portion ( to %10 size ). How can i do it in linux? Using something like par2 or ras. Both are apt-gettable and probably easily googleable. -- Didi kfir = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: grabbing video using bttv kernel module
Hi scipio, On Fri, Mar 04, 2005 at 02:52:12PM -0800, Nzer Zaidenberg wrote: hello. I am trying to grab video using either YES tv tuner or an analog camera. both display fine under windows. I am trying to use either a MATRIX frame grabber with BT878A or an unamed BT878 card. (The matrix card has a conexant Fusion 878A) the ptoblem is its only visiable as PAL-Nc and even then its black and white. everything is displayed in full color under windows. from what i red on the internet it appears that i should use modprobe tuner=xxx (xxx=???) I am using either xawtv or tvtime. I use vanilla fedora core 3 distro. does anybody know what i should use as xxx? or if i am even in the right direction? I personally use tuner type=5 bttv card=8 but I have a card different from yours (a bt848 based one). I think you'll either have to google for your exact card and see what people suggest or simply try all the possibilities. You can see the list in linuxkernelsource/Documentation/video4linux/bttv/CARDLIST (probably also googleable if you do not have kernel sources on your machine). Good luck, -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: USB Keydisk Configuration
On Fri, Mar 04, 2005 at 05:41:16PM +0200, Ira Abramov wrote: Quoting Ilya Konstantinov, from the post of Fri, 04 Mar: It all depends on whether you partitioned your disk-on-key. By default, you don't (just like you don't partition floppies). And there's nothing wrong about mounting the disk device itself (sda) instead of a partition on it (sda1). except if you want interoperability. most people get their DOK prepartitioned with part1 formatted as VFAT. this is how windows machines expect to see it too (it being any USB mass storage device, not to be confused with USB cdrom devices, which are never partitioned) That's also what I thought, until around a week ago. Then a student came with a DOK that wasn't partitioned, and Windows did see it. I wasn't in tau, but the people that looked at it said that Windows saw two drives - one a floppy (it was assigned B:) and the other an unpartitioned, formatted disk. In linux they saw only sda (and when mounting it, saw the contents of the disk part in Windows). So things are a little more complicated. If anyone sees a document that describes this in detail, and how to work with such DOKs in Linux, I would appreciate a pointer. Device Boot Start End Blocks Id System /dev/sda1 ? 3709864 404456684344761 69 Unknown Duh, it tries to interpret your vfat filesystem as if it was a MBR structure. more exactly, a partition has already started where the partition table should have been. Why more exactly? You said the same thing in different words. Especially that it's not that it _should_ have been, as I said. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: bad sectors on a logical storage device?
On Wed, Mar 02, 2005 at 01:49:53PM +0200, Ira Abramov wrote: Howdie people, [snip] so as you can see, I can understand and forgive FS errors (afterall the snapshot is taken while the main LUN is mounted) but I can't explain the I/O errors. My guesses are: * these are FS errors misreported as I/O errors (makes little sense) Maybe the NAS does not allow (or not easily) reading from snapshots blocks that were never written to? Does it have its own log? Anything interesting in it? Did you try reading from the disk directly (e.g. with dd)? Does it work? The only FS errors misreported as I/O errors I have seen are with corrupted FSes that had (bad) pointers to non-existing sectors. The sector numbers in your case are small so I do not think that's the case. * the kernel, although newer (2.4.18 IIRC) is still too old for this (odd?) I have no idea. My wild guess is that it's irrelevant of kernel version, as long as the driver is good and is compatible with the kernel. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: projector and Linux
On Thu, Feb 24, 2005 at 10:44:21AM +0200, Gabor Szabo wrote: Now that I know I should have checked the Linux support of that nice wireless card before I bought it (thanks Mark) I'll try to be more clever this time. I am about to buy a projector (you know, the one that I can use for presentations and trainings). While I think there should be no OS issue with it I try to cautious this time. So is there any Linux specific issue lurking out there that I should take care before buying? Specifically I got an offer of 6880 NIS for this projector: http://www.panasonic.co.uk/projector-mobile/ptlm1e/index.htm Will it work with a Linux notebook ? I am yet to see any problem with a specific _projector_, but we did have problems connecting projectors to some laptops running linux - that is, they (the card/driver) force only modes suitable for the builtin monitor, which look pretty bad on most projectors (which expect mostly vesa-compatible modes). Any other recommendations ? If you talk about a specific laptop, try it on that projector. If you can't, and you never connected it to a projector, try it on at least some projector. In addition, I do not recommend using the projector's dithering/smoothing, but rather change your resolution to its own (800x600 in your case) before your lecture and make sure it looks nice. If it doesn't, first try to fix the presentation, only then use the projector's smoothing. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: system clock loops
On Mon, Feb 21, 2005 at 12:48:01PM +0200, Tzafrir Cohen wrote: Hi I have a strange problem here: the system clock of my server keeps changing. The following is from the output of 'date' run from the same shell about 1 second apart: 12:13:37 13:25:08 12:13:34 As you can see, the clock occasionally loops back (it keeps in the range 12:13:34-38) and occasionally decides to move about 72 minutes forward. I have already eliminated (that is: killed) ntpd . 'ps auxww |grep ntp' shows nothing. Obviously anything that assumes a steady system clock misbehaves. Any idea what else may play with the system clock? Any cron jobs? -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Virtual fdisk?
On Sun, Feb 13, 2005 at 09:28:48PM +0200, Omer Zak wrote: When the partition table is destroyed (fully or partially) in hard disk, it needs to be reconstructed using patterns in the rest of the hard disk, documentation and shrewd guesses. After a guess is made about the extents of the partitions, the partition table can be edited and updated using fdisk. However, this must be done only after the correct partition sizes and positions have been figured out. Sometimes, this needs to be found by trial and error. It is very dangerous to perform the trial and error on the hard disk itself. However, sometimes it is not feasible to perform sector-by-sector copy (using dd) of the hard disk to another disk, due to its size. Therefore, it is desirable to have a way to instruct the OS to mount the hard disk in RO mode, and access its contents as if the partition table is such-and-such (rather than the partition table actually written into the hard disk). You can try something like losetup -o OFFSET /dev/hda /dev/loop0 mount ... /dev/loop0 ... You can't (as far as I know) set the length of the loop device, so you'll have to trust the filesystem's code to behave (most do). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: OT: A 24port switch HT-8016SR
On Thu, Feb 10, 2005 at 02:07:51AM +0200, Shachar Raindel wrote: I am feeling lucky in google gave this: http://www.hostnet.com.tw/ethernet%20switch/8016sr.htm Sounds like it has VLAN support (or at least an option for such), but Of course, that's what I said. But, you need to find out how to do this They do not tell you how to use it. I think I'll have to try and ping and tcpdump in several ports to find out. I did see some other company's product with similar toggles (but only 2, this one has 4) which has hard-coded VLANs according to the state of the toggles. I guess this one is the same, just a lot of work to find out exactly which port belongs to which VLAN on each state. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
OT: A 24port switch HT-8016SR
Hi all, Sorry for posting off-topic etc. I have a cheap 24port switch of model HT-8016SR. According to google it's made by a taiwanese company named hostnet. There are no manuals on their site (and I do not have one either). Of the only 3 results googles returns (not including very similar ones which are indeed very similar, I checked), 2 are from them and 1 is from Sivan Forums. I did not manage to read the actual messages, but it makes me think they might be common in Israel. That's why I ask here. The switch has four toggle switches marked TRUNK 1 2 3 4, which makes me hope I might be able to define VLANs on it. So my question is: Does anyone here have such a beast? Its docs? Any idea about it (or specifically what do these TRUNK buttons do)? As far as I can tell, it's unmanaged. Not through network nor a console port. Thanks, -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: smbfs - not mounting after windows reboot
On Tue, Feb 01, 2005 at 12:12:18PM +0200, Kfir Lavi wrote: Hi, i have linux and windows xp box. The linux is Debian Sarge. I have some mount points on the xp shared to the local network. After i restart linux, i can mount the shared on the xp. If i boot the xp, then i cann't mount and i get: 6931: protocol negotiation failed SMB connection failed Now if i will reboot the linux system, then i will be able again to mount. I guess smb is storing some data. I have searched /var for samba files, but didn't had success with those file, to nutralize them. Any insight? Not really - but two ideas: 1. Try doing the mount with strace. Both the first one (that works) and the second. 2. It might actually be a bug in the kernel. Try rmmod/insmod smbfs and see if it helps. I of course assume you did try to umount/mount and it did not work. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: CPU temperature
On Sat, Jan 29, 2005 at 03:23:39PM +0200, Ira Abramov wrote: Howdie people, I have a new machine running just under a month at the hosting farm at Nezeq Benleumi. it's a generic 1U rigged by 5 exits which I thought was well ventilated, however the following messages started appearing in the logs and on the console: Message from [EMAIL PROTECTED] at Sat Jan 29 15:03:22 2005 ... jenna kernel: CPU1: Temperature above threshold Message from [EMAIL PROTECTED] at Sat Jan 29 15:03:22 2005 ... jenna kernel: CPU0: Temperature above threshold Message from [EMAIL PROTECTED] at Sat Jan 29 15:03:22 2005 ... jenna kernel: CPU0: Running in modulated clock mode Message from [EMAIL PROTECTED] at Sat Jan 29 15:03:22 2005 ... jenna kernel: CPU1: Running in modulated clock mode (it's a hyperthreading P4, obviously) These seem to come directly from the CPU, not from your w83627hf. IIRC there are tools that can show the actual readings from the CPU itself, don't remember names. Try searching freshmeat for sensor, temperature, etc. As far as I know, lm-sensors never reads this. It only reads data from the relevant chip on the motherboard. searching around google tells me this is a cooling problem, but... (and there are several buts!) 1. nobody says if it is a big problem or a little problem. You can always stick a non-computerized thermometer there and see what the actual temperature is. Then look up the specs at Intel (I think a P4 shouldn't suffer much from up to around 60-70deg C) and decide how bad this is. 2. nobody says if this is the kernel just reporting me what happend to the CPU or the kernel deciding to switch the CPU to a modulated clock mode I also don't know. 3. I tend not to trust the way the kernel measures the CPU temp since the numbers I got from the output of sensors never made much sense, and the CPU is not under any serious stress in there. As I said, I do not think it's the same source. chopped out the memory module reports and fitting line length to a readable Email length, this is the output of sensors: w83627hf-isa-0290 Adapter: ISA adapter VCore 1: +0.02 V (min = +1.95 V, max = +2.16 V) VCore 2: +1.18 V (min = +1.95 V, max = +2.16 V) +3.3V: +3.41 V (min = +3.14 V, max = +3.47 V) +5V: +5.16 V (min = +4.76 V, max = +5.24 V) +12V: +12.28 V (min = +10.82 V, max = +13.19 V) -12V: +1.62 V (min = -13.18 V, max = -10.80 V) -5V: +0.43 V (min = -5.25 V, max = -4.75 V) V5SB: +5.64 V (min = +4.76 V, max = +5.24 V) VBat: +3.22 V (min = +2.40 V, max = +3.60 V) fan1:0 RPM (min = 675000 RPM, div = 2) fan2:0 RPM (min = 96428 RPM, div = 2) fan3: 7336 RPM (min = 1318 RPM, div = 8) temp1: +127 C (high =-118 C, hyst = +1 C) sensor = thermistor temp2:+116.5 C (high = +80 C, hyst =+75 C) sensor = diode ALARM temp3:+127.0 C (high = +80 C, hyst =+75 C) sensor = thermistor ALARM vid: +2.050 V (VRM Version 8.2) alarms: Chassis intrusion detection ALARM Are you sure the entirety of [lm-sensors modules, sensors program and libraries, /etc/sensors.conf] are of the same version and up-to-date? You should realize that the modules report the raw data they get from the sensors, and this gets translated to what you see here according to rules in /etc/sensors.conf. You should find the raw data somewhere under /sys. Try find /sys -iname *w83627* In a specific machine here, which is i2c - not isa, it's under /sys/bus/i2c/drivers/lm85/. A simple way to see how close to reality this is (or even send a fix to lm-sensors if needed), assuming you can have some downtime, is to look at the BIOS's report at various actual temperatures (e.g. using a good ventilator to cool everything a lot, look at both sensors and BIOS while it's getting warmer and finding the connection, which should be linear). -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: CS Languages for Teaching [was Re: Looking for a viable alternative to MS access.]
On Wed, Jan 26, 2005 at 10:25:58AM +0200, Ira Abramov wrote: Quoting Tzafrir Cohen, from the post of Wed, 26 Jan: On Tue, Jan 25, 2005 at 04:28:12PM +0200, Micha Feigin wrote: first also makes people use c++ as a functional language instead of as an OO language. Just to get the terminology right: I figure you meant procedural. every 9-12 months, this argument about the best first language comes up. half of the people arguing with opinions they won't gudge from and half trying to throw in half-knowledgable remarks to show they too exist, and never does anyone agree. so allow me to add to the tradition! Python, gentlemen! it can be OO or Procedural (and even pure functional I was once told). the syntax is clean, very little syntactic sugar, no odd compilersyntax for a newbie to learn, richer than Java, clearer than C and C++, and more widely used in the practical world than Pascal or LISP. And, may I add, has a nice, free book, called Learning with Python. Maybe not as deep as Structure and Interpretation ..., but not bad either. Now get ready to rumboll! Did you mean Rambo, or Rebol? I am sure you'll find people that will be in favour of Rebol as a first language. :-) -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: CS Languages for Teaching [was Re: Looking for a viable alternative to MS access.]
On Wed, Jan 26, 2005 at 06:04:30PM +0200, Micha Feigin wrote: On Wed, 26 Jan 2005 10:52:14 +0200 [EMAIL PROTECTED] (Yedidyah Bar-David) wrote: And, may I add, has a nice, free book, called Learning with Python. Maybe not as deep as Structure and Interpretation ..., but not bad either. Notice that Structure and Interpretation ... is not a lisp book, it used lisp as a tool. Will have to look at learning python though, always wondered if its going to be useful enough for me to spend the time learning, although I think that for my work I am stuck with matlab and c (really don't feel like learning fortran at this point ;-) Tt does have a secondary title which is How to Think Like a Computer Scientist. It does aim much lower than Structure etc., as I said. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Different res for different users
On Mon, Jan 24, 2005 at 06:04:35PM +0200, Shoshannah Forbes wrote: On 23/01/2005, at 23:36, Shlomi Fish wrote: I'm not sure that's the right place for it. You need root permissions to modify /etc/X11/XF86Config. Putting it there does not make sense on Linux. Since this topic came up anyway: I know gnome allows different users to have different screen resolution settings (on the same machine). Anybody know if there is a way to set that independently of gnome? The xorg.conf file seems rather system-wide, which is a strange thing for a multi-user system. Any pointers? Thanks! I do not know how gnome does it, but it probably uses the RANDR extension. You can probably do this yourself, but it's not built into the default configs of most distros I have seen. The main tool is a little program called xrandr. Read its manpage. Then create some kind of interface that allows the user to choose their favourite resolution, among those available (I know there is a GUI for the selection itself, don't recall its name - I think it's part of gnome - and you can use xrandr to see what is the current selection), keep it somewhere (e.g. $HOME/.myresolution), and patch the session starter (usually /etc/X11/xdm/Xsession or /etc/gdm/Xsession or something like that) to switch to this res before running the actual session program. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Different res for different users
On Tue, Jan 25, 2005 at 10:06:34PM +0200, Arnon Klein wrote: It's a funny coincidence, but my uncle (family tech support, you know...) asked me the same question today. Here's my script for doing it. If anyone cares to make something graphical from that, I'll be happy to learn from it... As I said earlier, there already is a GUI for xrandr. I was too lazy then to search it, but now you made me - apt-cache search xrandr says, among other things, gnome-randr-applet. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: Looking for a viable alternative to MS access.
On Mon, Jan 24, 2005 at 09:55:13AM +0200, Danny Lieberman wrote: Tazahi I've been following this thread and I wanted to shed a different light: I'm not a Microsoft advocate but lets get the facts straight: a. MS Access developer is free in Office Not exactly. It is built into Office Pro, but not in all Office bundles. b. The Academic licenses for Office are about 30% of the commercial MOLP which is about 20% of the single unit list price - so it costs peanuts for the Technion to develop in MS Access I agree it should not cost much. But they might have other considerations/needs. c. MS Access runtime (Jet) is free for unlimited distribution Jet, as far as I understand, isn't the runtime of MS Access. Actually, there is not such a separation between Runtime and Development time, in Access. You can't ship an executable+runtime. Jet, as far as I understand, is the runtime of Access's Database, that is, it's the code you need to read/write mdb files. So if the app is Access-based, you need Access. If it's e.g. written in VB, and only needs to rw mdb files, you can distribute it without limitations (plus Jet and VB runtime). d. You can use PHP on a Windows server to access the MDB database by ODBC since there is no limitation to using an MDB in a runtime server Indeed, but that does not help you if the app is Access. If you only need a DB, mdb might (or might not) be an acceptable choice. But he talked about Access is a development platform. If mdb is all you need, there are a few other small DBs. I played a bit with SQLite and liked it. I am not sure it does everything Jet does, but it's probably good enough for most such small projects. Second of all, with all due respect - this sort of smacks of i'll bite off my nose to spite my face (We dont use Access because ... ) MS Access is a widely used tool,and when your students go out looking for work they'll have a much better shot at a job with Access than with Ruby. Think of it like this - it's a tool and engineering schools like the Technion are supposed to teach you to find use tools. If the best tool for this job is a pocket knife thats the tool I would expect a Technion grad to use. Although I am from a CS school, which isn't exactly engineering, we do have here some discussions about what is our mission and what we should teach. Clearly we can't (and don't want to) teach only theoretic stuff. But I am also not sure we should choose our languages/platforms _only_ based on popularity and on increasing the number of relevant lines in our students' resume. If someone wants to program Access they do not need to spend 3 whole years here (or 4 in the Technion). Maybe Ruby isn't known enough. I am pretty sure, though, that PHP will serve them very well when looking for jobs. It's also probably more work than Access, so I understand Tzahi's request. To comment on Tzahi's original questions - I have no real experience with any of this, but I did see in the past some relevant projects based on PHP. Search freshmeat. Examples: bforms, bif, browse_edit, dblib, eleven, phpbuns, phpcodegenie, php_gen, phplabware, phplus, prado, webgen, yawp. They are from my .whatsnewfm.db.hot, in alphabetical order :-) I never used any of them. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]
Re: c++ sigmentation fault happens only in linux
On Mon, Jan 24, 2005 at 11:12:28AM +0200, Aviv Goll wrote: On Mon, 24 Jan 2005 05:16:23 +0200, Micha Feigin [EMAIL PROTECTED] wrote: At Mon, 24 Jan 2005 02:45:52 +0200, Aviv Goll wrote: hi, I'm currently writing an assignment in c++ using g++. according to some printouts, during the following lines: stringstream Fstr; fstrblah blah; Just tried running just that and had no problem. that's the whole point, I have the exect same lines (blah blah.. changed of course) written in about 10 other places in the code (in different methods, no colition with the decleration of Fstr), none of them causes segfault. and about the typo, it exists only in my Emails :-) the program says Fstr... Aviv, nobody thought that modern C++ compilers/OSes can't compile/run this piece of code. The bug is obviously elsewhere in your program. Now, traditionally, it would be hard to debug - you'd, as someone else suggested, find a minimal subset of your program that segfaults and debug it. But things changed since traditionally. Now there is valgrind. And you already know about it, as your first email showed us. Why won't you simply sit down and clean all the errors it talks about? Each and every one of them has huge chances to be a real, important, hard-to-find bug, that you got reported for free. If you had said My program segfaults, valgrind says nothing, now, that would really be strange. But I think anyone who read this thread up to now is completely sure, as I am now, that the segfault is a result of one of the already-found-by-valgrind errors. Simply fix them. -- Didi = To unsubscribe, send mail to [EMAIL PROTECTED] with the word unsubscribe in the message body, e.g., run the command echo unsubscribe | mail [EMAIL PROTECTED]