[freenet-support] Re: Automatic server retry of failing documents
Toad <[EMAIL PROTECTED]> writes: >> What is your definition of dirty hack? > > Fair enough. Would you say that the solution is to make sure the images > load the first time from IMG SRCs? If that were possible, sure. At least if they are available. If not (like edition headers), Freenet will not manage it to fetch it... It might at least try a bit longer. Usually images RNF in only a few seconds here... mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: Automatic server retry of failing documents
Toad <[EMAIL PROTECTED]> writes: > On Thu, Jun 17, 2004 at 06:33:22PM +0200, Michael Schierl wrote: >> Toad <[EMAIL PROTECTED]> writes: [putting every image into an iframe to work around Freenet's reloading bugs] >>> It works pretty well on the SSKvsCHK site :) >> >> One should not justify a dirty hack by "it works pretty well" since >> most dirty hacks will... > > Why is it a dirty hack? Do you think it is usual behaviour to put images into IFRAMEs? I have only seen that on Freenet. And since it is more work (and needs more resources on most browsers) and is only done to work around another problem, I usually call thinks like this "a dirty hack". So: - there is a usual solution (IMG tags for images). - it does not work in a specified environment (Freenet). - there is no interest to fix that solution in this environment, because - there is a "simple" way to work around that problem (IFRAME) - which uses techniques not appropriate for the problem (starting a whole HTML parser subinstance just for rendering an image) and - makes migration of existing work harder (mirroring a website into Freenet requires s/img/iframe/) ==> dirty hack. What is your definition of dirty hack? mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: Automatic server retry of failing documents
Toad <[EMAIL PROTECTED]> writes: > On Mon, May 10, 2004 at 10:17:30PM +0200, Michael Schierl wrote: >> Toad <[EMAIL PROTECTED]> writes: >> >> >> b) the failing file is an image. In that case it just disappears and >> >> you have to reload the page manually until you have all >> >> images. (alternatively, you can open a browser window/tab for every >> >> single image you want to have; IMHO this is no real solution either.) >> > >> > This is what IFRAME is for :) >> >> One iframe per image? IBTD. > > What's IBTD? I beg to differ? I thought it to be "I beg to disagree", but that is quite the same... > It works pretty well on the SSKvsCHK site :) One should not justify a dirty hack by "it works pretty well" since most dirty hacks will... mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: Automatic server retry of failing documents
Toad <[EMAIL PROTECTED]> writes: >> a) one does not use Fproxy for fetching a file > > In which case whatever you did use would retry. Not necessarily. Scripts talking FCP via netcat most likely won't... And, I think it is not an option that *if* every tool retried (just to make it work in b0rken freenet state) to "abuse" this by requiring tools to do so. (this argument is a bit like "why remove popups on my website? there are popup blockers anyway.") >> b) the failing file is an image. In that case it just disappears and >> you have to reload the page manually until you have all >> images. (alternatively, you can open a browser window/tab for every >> single image you want to have; IMHO this is no real solution either.) > > This is what IFRAME is for :) One iframe per image? IBTD. mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: Automatic server retry of failing documents
Roger Hayter <[EMAIL PROTECTED]> writes: > Michael Schierl <[EMAIL PROTECTED]> writes >>For me it seems as Fred is too fail-fast. It may try for a few more >>seconds or minutes instead of returning the RNF nearly immediately >>(and very often). Putting that retry to the user's side (being it the >>browser or a FCP app) is no good idea IMHO. >> > How would you distinguish messages you don't want very much (like > possibly non-existent Frost messages) from ones you do want a lot? By not using Frost? If that is really an issue, one could specify an "importance" flag in FCP. But IMHO that many RNFs are a bug in Fred. > It > is likely to generate quite a lot of extra traffic to automatically > retry all RNFs till you get something, even if the user has gone on to > something else. Not "until you get something", but for some more time. I guess most users will not change their mind if they cannot get a site within a second (which is usual time till RNF for me ATM). mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: Automatic server retry of failing documents
"Niklas Bergh" <[EMAIL PROTECTED]> writes: >> Now I believe I can live with freenet being slow and that >> many documents are not immediately available. What is >> annoying me is that _I_ will have to do the retrying. Why is >> that not a task for the server? > > You sure? I thought we had a meta-refresh tag on those RNF/DNF pages? > Wont the browser automatically retry the page after a while if you leave > it to? Hi again, wanted to try Freenet again ;-) - seems to work quite well. However, those RNFs are really a PITA. Yes, those RNF pages have meta refresh tags on them; however, that does not help if a) one does not use Fproxy for fetching a file b) the failing file is an image. In that case it just disappears and you have to reload the page manually until you have all images. (alternatively, you can open a browser window/tab for every single image you want to have; IMHO this is no real solution either.) For me it seems as Fred is too fail-fast. It may try for a few more seconds or minutes instead of returning the RNF nearly immediately (and very often). Putting that retry to the user's side (being it the browser or a FCP app) is no good idea IMHO. mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: The network is busy, are you even connected to the Internet?
Pierre Abbat <[EMAIL PROTECTED]> writes: > I keep getting errors like this when browsing Freenet: > > The network is busy, please try again later. > > Couldn't connect to the network. Are you sure you have configured Freenet > correctly? Also make sure that you are connected to the internet. > > A tcpdump shows that I am indeed connected to the Internet, and there is > plenty of slack bandwidth. I can get some Freenet sites. What could the > problem be? Switch the web interface into expert mode and you will get a more helpful (for developers) error message. Michael ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: unable to build map file
"Lloyd Burris" <[EMAIL PROTECTED]> writes: > When ever I try to submit a freesight at the end fiw says unable to > build a map file. That is most likely a FIW bug (since *building* mapfiles has nothing to do with Freenet, only inserting them). Please post some more information (contents of the report-log and logfile.log) to the FIW mailing list. Thanks. You might also check that the main page for the freesite is named "index.html" (lowercase) - there is some bug in current FIW version that makes trouble if there is a file "Index.html" or "INDEX.HTML". mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: What does a map-file look like?
"Garb" <[EMAIL PROTECTED]> writes: > You are right - that certainly seems to be exactly what I am looking for (if > we exclude J.Lo's private phone number...). But where is that information > actually located? It is not in the HTML-source of the page. Could it be in a > special file with a name that FIW can somehow deduce from the URI of the > main page? In Freenet there is no such thing as a "file" (surely, in your datastore there are files, but there needn't be. E.g. Freenet 0.4 used one large file as datastore). Every key in Freenet can have 2 different portions of data. Metadata and Raw Data. While tools like FProxy and FUQID usually show raw data to you (since metadata is for internal purpose) that does not mean there is no metadata saved. Another example: In modern OSes (including Windows NT, 2k, XP) you can add "file permissions" to a file. They are not saved inside that file (you'll notice that when you look inside before and after you edit it) and they are not saved in some "special" file either - they are simply saved in the file system (where file names are saved as well). Or, on XP, you can even assign "real" metadata to a file, like an author name, which will not end up inside that file. (If you copy that file onto a CD-ROM or floppy disk and then copy it back, the author entry will disappear). And similar to these kinds of metadata in file systems, Freenet has things like that as well - it is just used for different purposes. If that helps you, just think of two "files" behind every freenet key - one for metadata and one for data. (Actually this is wrong, since both metadata and data of one key will always be saved on the same Freenet node, while two different keys don't need to) Michael ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: What does a map-file look like?
"Garb" <[EMAIL PROTECTED]> writes: > I have been using FIW for inserting sites, but because I'm a hands-on kinda > guy and a masochist, I would really like to try doing it a bit more > manually. Unfortunately I don't seem to be quite smart enough to figure out > how. My problem is the contruction of clickable links to files that are > already in freenet. How do I do that? just link to the key, preceded by a slash, e.g. Maybe the GPL What know I A nice picture A nice picture with a nice name > Then I heard something about a "MAP"-file, which I am lead to assume could > be the solution to all my problems - or at least the ones not related to > girls or money. It appears to be some sort of cross-reference between the > real-world file-names and the URIs, but I have thus far been unable to find > out exactly what it looks like or how to make them. And I'm still not > certain, how that would solve my problem with telling the browser about > fred's location. FIW will build a mapfile when you insert your site so that you can use "nice" names for files inside it. You could tweak this file by adding stuff into your .fiw/mapfile.ini inside the project directory, but I think that is overkill for linking to foreign sites. It might be useful if you want to give files an address below your SSK (to prove others that you approve its contents), but is not needed for linking to foreign files. To view the "raw mapfile" of a freesite, open FIW's metadata browser and enter the URI of that site up to the first // in it. (You might see a DateRedirect first if that site is DBR, but just move your cursor on that DateRedirect line and hit Return to view the current mapfile behind it.) The format for mapfile.ini is a bit more "cooked" than the mapfiles inside freenet, e.g. [EMAIL PROTECTED] Then you can directly link to picture.png from your site and viewers won't be able to notice that that is a file you didn't insert yourself. > Any of you brains out there have an idea or a pointer as to where I could > look for info? If you are more interested about what is happening "behind the scenes", try to get the "Nubile" or "Freenet Explained" freesites. AFAIK they are archived somewhere outside Freenet as well, but I don't remember the URL. mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: slowdom in freeville
Toad <[EMAIL PROTECTED]> writes: >> So, this bug is really somewhere inside freenet. Either the accept() >> call was not handled quick enough (dunno how accept is called in nio) >> or Fred really closes open sockets without having sent any data. > > Consistently? Can you write me a test case? Since it has never happened on my node, no; sorry. I could try to write an app that just sets up a bunch of ClientPut requests, but i cannot be sure if they are the reason for the disconnection of fred. BTW i don't have any node atm anyway (my build on hd is 6441 and it hasn't been started for a few months), so i'd have to set up a new node first... Perhaps I'll do that. But not *now*. BTW2 have you fixed that bug where a plain site was detected as a container (i mailed you on 2004-01-10 with a way to reproduce it) and thus unbrowsable? mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: slowdom in freeville
Michael Schierl <[EMAIL PROTECTED]> writes: > However, after a few tests (replaced Fred by a simple prog that says > nothing but disconnects on a keypress) i can see that I get an > IOException like the one above when the connection is closed from the > peer, but a ConnectException when I run into a SocketTimeout. Seems to > be some work for me to fix that (in that case it should not mess up > the log file)... St00pid me... The connect exception was caused by my test app which closed the socket and since FIW reconnects at once, it gave an error then... The SocketTimeoutException occurring first is (as expected) not written to the error log file. So, this bug is really somewhere inside freenet. Either the accept() call was not handled quick enough (dunno how accept is called in nio) or Fred really closes open sockets without having sent any data. mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: slowdom in freeville
Toad <[EMAIL PROTECTED]> writes: >> > I also get quite some of these: >> > >> > java.io.IOException: Premature end of stream >> >at fiw.fcp.FCPMessage.readMessage(FCPMessage.java:34) >> >at fiw.fcp.FCPConn.insertStream(FCPConn.java:263) >> >> Hmm. Why does fred close his connections prematurely? It's not because >> you are restarting fred? > > I don't think it does. It's probably a socket timeout on FIW's end. Hmm. That should give SocketTimeoutException (which is caught properly), shouldn't it? Does not seem so. However, after a few tests (replaced Fred by a simple prog that says nothing but disconnects on a keypress) i can see that I get an IOException like the one above when the connection is closed from the peer, but a ConnectException when I run into a SocketTimeout. Seems to be some work for me to fix that (in that case it should not mess up the log file)... mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: slowdom in freeville
Someone <[EMAIL PROTECTED]> writes: > I get many of these (using FIW 0.08): > > #11 02 ***RouteNF *removed* (Chunk 6) [42/42/42] > > they can't be right as I only have 50 nodes in my routing table and a max > connections number of 100. Hehe. That 42/42/42 (do you know Douglas Adam's "Hitchhiker's guide to the galaxy"?) is just a code for "bad things happend" as any other combination of at least twice 42. In that case it means that an insert returned a DNF message which is invalid according to the spec and thus treated as a RNF. (RFC1122 Robustness principle). Toad told me that he had fixed that but most likely he hasn't... > I also get quite some of these: > > java.io.IOException: Premature end of stream > at fiw.fcp.FCPMessage.readMessage(FCPMessage.java:34) > at fiw.fcp.FCPConn.insertStream(FCPConn.java:263) Hmm. Why does fred close his connections prematurely? It's not because you are restarting fred? > It also complains very often that it can't fetch an inserted chunk, while I > can fetch it without a problem through fproxy. This leads to continous insert > retries whitout FIW inserting the mapfile. set the "default priority" in advanced settings so that the mapfile is inserted "first". This should speed your insert up somehow. Then you can disable to "ignore local datastore". This will make your insert much faster, but much less reliable as well. I don't know what FUQID is doing, but if it does similar things, I understand that insert is faster. Greets mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] Re: slowdom in freeville
Someone <[EMAIL PROTECTED]> writes: > sites. Maybe because of the still very slow insert performance of fiw, which > still takes 36 hours to successfully insert a 4 MB freesite (fuqid inserts much > faster). Any ideas why? I'd like to make FIW fast too ;) mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support Unsubscribe at http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support Or mailto:[EMAIL PROTECTED]
[freenet-support] New mailing list for FIW!
Hi guys, You might have noticed that i do not have the time to follow all discussions on the freenet mailing lists. However, I do not want to give up FIW development (although the last update has been some time ago). So I decided to create a discussion mailing list for FIW - additionally to the CVS mailing list that already exists. You can find information about this new list at https://lists.sourceforge.net/lists/listinfo/fiwiz-devl Additionally, I added a "snapshots" dir to the sf.net homepage (http://fiwiz.sf.net/snapshots/ where I can provide latest CVS builds for those who like to be on the bleeding edge. So, for additional discussions (and support questions as well) please use the new mailing list. Some information about the current FIW roadmap: Thelema has committed a new FCP layer a week ago for FIW which should make it more robust against Freenet "issues". Today i managed to spot the last large bug in the FEC part, so I hope it is "stable enough" now. If you want to test it, try current CVS (committed a few minutes ago, so it might not be available on public cvs yet) or the current snapshot. I also added kinda "test case" for the fcp layer - it connects to freenet and tries to do some usual things like generating keys, inserting and fetching keys (at htl 0) etc. FEC is not covered by it yet. To try that, fetch the source from the src.zip inside the snapshot or from cvs and do a "ant fcptest"; if you find a (recent) freenet build where this test creates errors (not only warnings), please tell me. Now most code uses deprecated methods in the old FCP layer (that used static methods...) which call the new fcp layer. I'll try to clean this up asap and then (if there is still time left) add some new shiny SWING gui (additionally to the AWT GUI) to it. Whether the new gui comes or not - i plan to release 0.08 (or 0.08-pre1) in mid-january. I'll announce it on the FIW mailing list then. if you have suggestions or feature requests, you know how you can reach me, mihi ___ Support mailing list [EMAIL PROTECTED] http://news.gmane.org/gmane.network.freenet.support
[freenet-support] Re: 5032 & FIW 0.07 - FEC Encoding problem?
On Thu, 20 Nov 2003 12:34:11 +0100, [EMAIL PROTECTED] wrote: > this is a FIW issue which is known for some time now but has never been fixed I know it since a few days after i released FIW 0.07. And i fixed it in CVS already. So if you can use CVS and compile it yourself, give it a try. Otherwise: setting sotimeout to zero should "fix" the problem as well. BTW: thelema is trying to rewrite the FCP layer atm (AFAIK), so I won't release a new version before that is done (This might fix this bug as a side effect as well...). mihi ___ Support mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support
[freenet-support] Re: Website insertion ?
On 15 Nov 2003 22:13:21 -, panta-admin wrote: > Michael if you read this could you please invest some of your precious time > into fiw, it cant handle splitfiles at the moment, if you read mxbee's message in devl: fuqid cannot atm either - so it is a freenet bug. sorry. if you already tried fiw, you could still try fishtools. or insert splitfiles as CHK@ with the web interface and use mapfile.ini to reference them in your site. mihi ___ Support mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support
[freenet-support] Re: A few basic questions
On Tue, 28 Oct 2003 23:34:21 -0500, Nicholas Sturm wrote: >> >> b/w > > As a photographer that means "black and white" > > On the Internet it often means "by the way" > > What does it mean here? bandwidth? mihi ___ Support mailing list [EMAIL PROTECTED] http://dodo.freenetproject.org/cgi-bin/mailman/listinfo/support
[freenet-support] Re: freesite for Frost FEC
[EMAIL PROTECTED] schrieb: > I'm using FIW, a swing application, with Kaffe 1.0.7, with only some > warning messages in console JFTR: FIW 0.04b is not (yet) Swing, it's still AWT. thx, mihi, not subscribed to support@, but hoping that it'll get through ___ support mailing list [EMAIL PROTECTED] http://hawk.freenetproject.org/cgi-bin/mailman/listinfo/support