Re: busybox update
On 31/08/07, Jérémy Bobbio [EMAIL PROTECTED] wrote: On Fri, Aug 31, 2007 at 01:45:24PM -0300, Otavio Salvador wrote: Eddy Petrișor [EMAIL PROTECTED] writes: Bastian Blank wrote: Hi folks I prepared a busybox update. The following changes are not yet done: - Remove ifconfig. Scheduled since at least 3 years. 3 users left. ppp-udeb being one of them (i.e. have you counted here ppp-udeb)? So this means broken PPPoE installs? ppp-udeb uses ifconfig in its postinst. We might start work to remove its dependency and use ip if possible. I took a quick look at it, attached is a possible (untested) patch removing ppp-udeb dependency on ifconfig. I don't have the time, energy or motivation to do the tests myself, but the changes look ok. I think is best to submit this patch in the bts (if didn't already). One can test himself if the thing works by doing what I did myself back in January[1]. If someone has time to prepare a custom image with this patch included I might me able to do it. [1] http://ramblingfoo.blogspot.com/2007/01/howto-test-pppoe-without-having-pppoe.html -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Bug#422713: lspci does not list anything (etch and dailies)
Package: installation-report Severity: important Hello, I tried the Etch MA image and an Etch netboot on a desktop. The installer failed to detect the network interface. I think the problem lies somewhere in the kernel or in some basic library since lspci did not returned any results in DI and also in the installed system. I tried also the dailies knowing that they use the 2.6.20 kernel, same results. I tried a ubuntu warty image to get the lspci output (I managed to, but I couldn't get that data out of the environment since the network card was not enabled); now I just tried a knoppix 3.4 image and managed to get the lspci output which is attached. (The old images were the ones I had at hand and I wanted to test a kernel prior to 2.6.18). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein lspci Description: Binary data
ppp_2.4.4rel-5 patch
Package: ppp
Tags: patch d-i
Severity: important
Hello,
Attached is a patch that closes a few open issues with the ppp-udeb.
The most important are #402450 (not bringing up the interface after
install and not saving a correspondent section in the target
/etc/network/interfaces) and the handling of bad login information
being typed.
Here is the changelog:
+ [ Eddy Petrișor ]
+ * added ${misc:Depends} to ppp-udeb
+ * removed debconf as a dependency of ppp-udeb since such a package does not
+exist in d-i environment
+ * ppp-udeb: the interface is now raised in the installed system
+(Closes: #402450)
+ * ppp-udeb: loopback interface snippet is added too, to make sure we have it
+in the installed system
+ * ppp-udeb: added error handling if the wrong login info is sent or other
+error is returned while calling the provider
+ * Added myself to Uploades with maintainer's consent.
Issues still open:
- hostname is not set (I think I can force netcfg to run the mandatory
steps, but I didn't had the time to do it during the winter holidays)
and does not yet blend well with netcfg
- finish-install.d was not working at some point in the past, but the
current situation is uncertain
Note that, in spite of the present issues, this release does not bring
any regressions, on the contrary, it just fixes existing issues.
--
Regards,
EddyP
=
Imagination is more important than knowledge A.Einstein
diff -ruN official/ppp-2.4.4rel/debian/changelog ppp-eddyp-dev/ppp-2.4.4rel/debian/changelog
--- official/ppp-2.4.4rel/debian/changelog 2007-01-09 19:47:49.0 +0200
+++ ppp-eddyp-dev/ppp-2.4.4rel/debian/changelog 2007-01-09 19:43:02.0 +0200
@@ -1,3 +1,19 @@
+ppp (2.4.4rel-5) unstable; urgency=low
+
+ [ Eddy PetriÈor ]
+ * added ${misc:Depends} to ppp-udeb
+ * removed debconf as a dependency of ppp-udeb since such a package does not
+exist in d-i environment
+ * ppp-udeb: the interface is now raised in the installed system
+(Closes: #402450)
+ * ppp-udeb: loopback interface snippet is added too, to make sure we have it
+in the installed system
+ * ppp-udeb: added error handling if the wrong login info is sent or other
+error is returned while calling the provider
+ * Added myself to Uploades with maintainer's consent.
+
+ -- Eddy PetriÈor [EMAIL PROTECTED] Sat, 23 Dec 2006 15:27:43 +0200
+
ppp (2.4.4rel-4.1) unstable; urgency=low
* Non-maintainer upload with maintainer's consent.
diff -ruN official/ppp-2.4.4rel/debian/control ppp-eddyp-dev/ppp-2.4.4rel/debian/control
--- official/ppp-2.4.4rel/debian/control 2007-01-09 19:47:49.0 +0200
+++ ppp-eddyp-dev/ppp-2.4.4rel/debian/control 2007-01-09 19:43:02.0 +0200
@@ -2,6 +2,7 @@
Section: admin
Priority: optional
Maintainer: Marco d'Itri [EMAIL PROTECTED]
+Uploaders: Eddy PetriÈor [EMAIL PROTECTED]
Build-Depends: debhelper ( 4.1.17), libpcap0.8-dev, libpam0g-dev, zlib1g-dev
Standards-Version: 3.7.2.2
@@ -24,7 +25,7 @@
XC-Package-Type: udeb
Section: debian-installer
Architecture: any
-Depends: ${shlibs:Depends}, ethernet-card-detection, ppp-modules, di-utils, debconf
+Depends: ${shlibs:Depends}, ${misc:Depends}, ethernet-card-detection, ppp-modules, di-utils
Provides: configured-network
XB-Installer-Menu-Item: 17
Description: Point-to-Point Protocol (PPP) daemon
diff -ruN official/ppp-2.4.4rel/debian/ppp-udeb.postinst ppp-eddyp-dev/ppp-2.4.4rel/debian/ppp-udeb.postinst
--- official/ppp-2.4.4rel/debian/ppp-udeb.postinst 2007-01-09 19:47:49.0 +0200
+++ ppp-eddyp-dev/ppp-2.4.4rel/debian/ppp-udeb.postinst 2007-01-09 19:43:02.0 +0200
@@ -207,7 +207,20 @@
log-output -t depmod
log-output -t ppp-udeb modprobe pppoe
-log-output -t ppp-udeb pppd call provider || true
+RET=
+log-output -t ppp-udeb pppd call provider || RET=$?
+
+if [ 0$RET -eq 19 ]; then
+ fail wrong login info detected
+ db_input critical ppp/wrong_login || true
+ db_go || true
+ exit 1
+elif [ 0$RET -ge 1 ]; then
+ fail unhandled error detected
+ db_input critical ppp/unhanled || true
+ db_go || true
+ exit 1
+fi
log-output apt-install ppp || true
diff -ruN official/ppp-2.4.4rel/debian/ppp-udeb.templates ppp-eddyp-dev/ppp-2.4.4rel/debian/ppp-udeb.templates
--- official/ppp-2.4.4rel/debian/ppp-udeb.templates 2007-01-09 19:47:49.0 +0200
+++ ppp-eddyp-dev/ppp-2.4.4rel/debian/ppp-udeb.templates 2007-01-09 19:43:02.0 +0200
@@ -47,3 +47,24 @@
Template: ppp/detect_progress
Type: progress
_Description: Searching for concentrators on ${IFACE}...
+
+Template: ppp/wrong_login
+Type: error
+_Description: Failed authentication
+ A PPPoE connection was attempted, but the provided login information was
+ incorrect. Because of this the authentication failed.
+ .
+ Please make sure you type the correct username and password.
+ .
+ The configuration of PPPoE is aborted. It can be attempted again by
+ selecting the relevant menu entry.
+
+Template:
Re: Bug#385150: Tests with PPPoE connection
tag 384121 patch tag 385150 patch thanks On 27/09/06, Gregory Colpart [EMAIL PROTECTED] wrote: On Tue, Sep 19, 2006 at 02:34:57PM +0300, Eddy Petri??or wrote: Please do test without the || true code attached to apt-install and with the db_stop line removed or commented out. I expect this to be the correct fix, and, as explained above, the || true being a lucky way out in some cases. Result is red debconf screen because failure. See _complete_ log here: http://gcolpart.evolix.net/debian/d-i/27092006/syslog Yeap, that is correct :-D . Ok, I looked over the apt-install code and saw that apt-install will return 1 is the package is queued fr installation, a thing which is perfectly acceptable for ppp :-D. I have attached a patch that should be applied over the current source ppp package and should fix both #385150 and #384121. I attached also the resulted ppp-udeb.postinst file after the patching (which should improve testing efficiency, since it can be transfered to the test machine via netcat). People who wish to test the new changes can grab the postinst attached and copy it via netcat (nc) in the installation environment (still the plugin moving is needed) over the original one. If you have other tests with PPPoE/d-i, say me. I know that last time I tested the package, the installation on the target system was not good enough and after the first reboot, although the ppp package was installed, the pppoe connection was not working. I _know_ this may be impossible for you to test, since you might not have a spare machine or a spare partition, so I won't mind if you can't do it. But, please, tell me if you can't do this test. Indeed I have no spare partition to test complete d-i with PPPoE and first reboot. I will *try* to have a new disk to test it next week. Thanks a lot for that. P.S.: The current ppp-udeb todo list is also attached, it would be nice to have it in the ppp source package. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein ppp-udeb_more_fixes.patch Description: Binary data ppp-udeb.postinst Description: Binary data ppp-udeb.todo Description: Binary data
Re: Bug#342053: DirectFrameBuffer crashes on PPC systems if HW accelerated drivers are used
On 25/09/06, Eddy Petrişor [EMAIL PROTECTED] wrote: A round of PPC tests would be useful, especially it happens to find owners of ATI or NVIDIA boards. Got one. Eddy, do you have the chance to test if forcing off DFB's HW acceleration makes g-i run on your Mac? I did, (Sven knows), it didn't work for me. I added by hand in the directfbrc file the option. Neither did video=ofonly At Sven's suggestion, I have ran the installer with disable-module=radeon in directfbrc. This worked (well, only if the linux_input module is deactivated, too). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: Bug#342053: DirectFrameBuffer crashes on PPC systems if HW accelerated drivers are used
On 23/09/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: Eddy Petrişor wrote: snip/ A round of PPC tests would be useful, especially it happens to find owners of ATI or NVIDIA boards. Got one. Eddy, do you have the chance to test if forcing off DFB's HW acceleration makes g-i run on your Mac? I did, (Sven knows), it didn't work for me. I added by hand in the directfbrc file the option. Neither did video=ofonly -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Bug#342053: DirectFrameBuffer crashes on PPC systems if HW accelerated drivers are used
On 23/09/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: Also, about the console font corruption with radeonfb, i would be interested in feedback of if it is a powerpc only issue, or ppc stuff ? No idea, i on no radeon boards :( $ lspci | grep ATI :00:10.0 VGA compatible controller: ATI Technologies Inc RV350 [Mobility Radeon 9600 M10] Any chanche to test if disabling HW acceleration also makes the g-i usable on machines equippped with ATI or NVIDIA graphic boards ( where atyfb and nvidiafb modules would be used in the case HW acceleration was not forced off ) ? Nope, but as soon as the fixed rootskel-gtk is uploaded, we can issue a call for testers on debian-powerpc, using the daily builds. A round of PPC tests would be useful, especially it happens to find owners of ATI or NVIDIA boards. Got one. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Bug#385150: Tests with PPPoE connection
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Gregory Colpart wrote: Hi, On Sat, Sep 16, 2006 at 03:01:43AM +0300, Eddy Petri??or wrote: It seems (thanks again Joey) that the problem is the db_stop line which, if commented out should kill the problem appear in both of these cases. Gregory knows this and said he will try to test with this change, but Abel should know about it, too. I would be really happy to find out that both issues disappeared when commenting out db_stop (it appears that I shouldn't do that in D-I). I use latest daily build for my PPPoE tests. Note that I use mv /usr/lib/pppd/2.4.4b1 /usr/lib/pppd/2.4.4 workaround for all my tests. I was expecting that :-) since no release of the ppp package has been made :-). Probably Marco is waiting for a fix for the issue we are talking about, too. * If I try without changing anything, I have red debconf screen. You have a full syslog here: http://gcolpart.evolix.net/debian/d-i/16092006/full-syslog.no-changes.failed * If I try with adding '|| true' in 'apt-install ppp' line, yeah, all is right !! After my PPPoE connection is up, I return to menu with no error. You have full syslog with -e and -x for debug: http://gcolpart.evolix.net/debian/d-i/16092006/full-syslog.true-patch.success.set-e http://gcolpart.evolix.net/debian/d-i/16092006/full-syslog.true-patch.success.set-x [Writing this mail offline, so I'll have to check the logs later] AFAIUI, debconf should never be ordered to stop, while in D-I (Debian Installer) environment, so I expect that the || true code is just helping the debconf protocol to get out of a messy situation, in some rare cases. Also, the 'apt-install ppp code should not fail; failing would mean that the installed system would have issues since its ppp package would not install. Needless to say that you don't want that ;-) . I don't try to test without db_stop because it seems to me that my tests above are sufficient. I don't know exactly why, but it works fine with the last d-i. Please do test without the || true code attached to apt-install and with the db_stop line removed or commented out. I expect this to be the correct fix, and, as explained above, the || true being a lucky way out in some cases. If you have other tests with PPPoE/d-i, say me. I know that last time I tested the package, the installation on the target system was not good enough and after the first reboot, although the ppp package was installed, the pppoe connection was not working. I _know_ this may be impossible for you to test, since you might not have a spare machine or a spare partition, so I won't mind if you can't do it. But, please, tell me if you can't do this test. Thanks a lot for your help. - -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.5 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFFD9VPY8Chqv3NRNoRAghXAKCwGR48+WP5lPVg/q28WaxDQHTQowCbBxUj SnzjblXVntwOizuQDoYFXWE= =nL8M -END PGP SIGNATURE- -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [g-i] Am I crazy or has linehight increased dramatically recently?
On 18/09/06, Frans Pop [EMAIL PROTECTED] wrote: Well, let's assume for now that I am not crazy. If I run g-i in vmware, I get this: http://people.debian.org/~fjp/d-i/g-i_lineheight-and-ugly-font.png That is almost double lineheight between choices! Also, does the font look uglier than it used to? Let's go on with bad things :-D Have you seen the vertical line in the upper right corner? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [g-i] Am I crazy or has linehight increased dramatically recently?
On 18/09/06, Frans Pop [EMAIL PROTECTED] wrote: On Monday 18 September 2006 16:31, Eddy Petrişor wrote: Have you seen the vertical line in the upper right corner? Yes, I already reported that one a long time ago... Hmm, I was under the impresion that the vertical line should have been fixed at the same time as horizontal issue was fixed. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: Bug#385150: Tests with PPPoE connection
On 05/09/06, Gregory Colpart [EMAIL PROTECTED] wrote: Hi, Here are bits from my (nightly) PPPoE tests with d-i. On Sat, Sep 02, 2006 at 03:06:45AM +0300, Eddy Petri??or wrote: Then tell me if things improve? With logs, if not. Same problem with debconf error. I retry/retry/retry with apt-install ppp || true patch and I was wrong, it is not the same problem (grrr, probably too tired during my last tests). Then I don't have red debconf screen but I return to Debian installer main menu. It sounds good but there are little problems with this menu. I go directly on a blank line between 'Install the base system' and 'Select and install software' choices. There is also another blank line between 'Partition disks' and 'Configure the clock'. Logs are here: http://gcolpart.evolix.net/debian/d-i/syslog-for-pppoe.3 (in this logs, I try to select a strange OK choice on the top of menu...) It seems (thanks again Joey) that the problem is the db_stop line which, if commented out should kill the problem appear in both of these cases. Gregory knows this and said he will try to test with this change, but Abel should know about it, too. I would be really happy to find out that both issues disappeared when commenting out db_stop (it appears that I shouldn't do that in D-I). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [g-i] new screenshots available (dejavu 2.9)
On 11/09/06, Davide Viti [EMAIL PROTECTED] wrote: the screenshots can be found in [2], and in case you wanted to compare them to the previous, see [3] feedback is very welcome [1] http://lists.debian.org/debian-boot/2006/06/msg01178.html [2] http://d-i.alioth.debian.org/gtk-frontend/screenshots/20060911_dejavu2.9/ [3] http://d-i.alioth.debian.org/gtk-frontend/screenshots/20060621_dejavu2.7/ The font size is different in the screenshots and, thus I can't tell if the half of breve issue has gone. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: RFC: countrychooser and IP location
On 10/09/06, Joey Hess [EMAIL PROTECTED] wrote: Robert Millan wrote: +iplocated_country=`wget -q http://api.hostip.info/country.php -O - | head -c 2` This is a third-party point of failure. Altough your code does gracefully degrade if it fails. Would Debian be able to run this service itself if we decided to use this? Or would that be a bad idea? Just curious. Wasn't geoip[1][ supposed to be something which did some location - IP association? Anyway, I rather like the idea of doing IP location, but I lean more toward only using the info to override the user's country in mirror selection, rather than overriding the whole countrychooser. For one thing, it wouldn't need to ask the user an insulting question (do you know what country you're in?) if the iplocated country was only used to change the default mirror. [1] http://packages.debian.org/unstable/net/geoip-bin -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Tests with PPPoE connection
On 01/09/06, Gregory Colpart [EMAIL PROTECTED] wrote: On Thu, Aug 31, 2006 at 04:21:33AM +0300, Eddy Petri??or wrote: It should be in /var/lib/dpkg/ppp-udeb.postinst, sorry for forgetting that info. /var/lib/dpkg/info/ppp-udeb.postinst ;) Indeed. Then tell me if things improve? With logs, if not. Same problem with debconf error. Logs are here: http://gcolpart.evolix.net/debian/d-i/syslog-for-pppoe.2 Still I have no clue... unfortunately. I am definetly now shooting in the dark: do that previos modification and also add set -x at the beginig of the script, after the #!/bin/sh line, and do the buggo dance again. Abel, it would be a good idea for you to do this set -x thing in your script, too. I think in your case it would be a lot more helpful. Someone suggested me to set a local pppoe server so I can test. If I don't understand what is happening I will try to set one and test (I never did this before, so I don't know what to expect/how hard it will be). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Tests with PPPoE connection
On 31/08/06, Gregory Colpart [EMAIL PROTECTED] wrote: Hi, Hello, Sorry for the lag of my answer, I was on pseudo-[VAC]. NP, this debugging is not such a pleasure :-D On Thu, Aug 24, 2006 at 03:24:31PM +0300, Eddy Petri??or wrote: Could you please set DEBCONF_DEBUG=5 at boot time and run the PPPoE step (after doing the move of the plugin) and send the relevant log section? Relevant means from where it says the PPPoE manu was selected until the place where it gives the errorwith multiselect backup. I retry with same ISO. PPPoE connection is OK but I have always debconf error. Here are logs: http://gcolpart.evolix.net/debian/d-i/syslog-for-pppoe Interesting part is probably: Aug 31 00:24:00 pppd[11082]: pppd 2.4.4 started by (unknown), uid 0 Aug 31 00:24:00 apt-install: Queueing package ppp for later installation Aug 31 00:24:00 main-menu[2142]: (process:10955): 0 Aug 31 00:24:00 main-menu[2142]: WARNING **: Configuring 'ppp-udeb' failed with error code 1 This is the same error Abel faces Hmm, if this is because of ppp-udeb, then it could be because of the apt-install command (shooting in the dark, but its worth a try). Abel, Gregory, could you try to modify (nano is available), before running the PPPoE menu item, the ppp-udeb.postinst script (at the end) so it reads: snip # FIXME - how can this be improved? [ -e /etc/resolv.conf ] || ln -s /etc/ppp/resolv.conf /etc/resolv.conf log-output apt-install ppp || true #DEBHELPER# /snip instead of snip # FIXME - how can this be improved? [ -e /etc/resolv.conf ] || ln -s /etc/ppp/resolv.conf /etc/resolv.conf log-output apt-install ppp #DEBHELPER# exit 0 /snip Then tell me if things improve? With logs, if not. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Tests with PPPoE connection
On 31/08/06, Eddy Petrişor [EMAIL PROTECTED] wrote: Aug 31 00:24:00 pppd[11082]: pppd 2.4.4 started by (unknown), uid 0 Aug 31 00:24:00 apt-install: Queueing package ppp for later installation Aug 31 00:24:00 main-menu[2142]: (process:10955): 0 Aug 31 00:24:00 main-menu[2142]: WARNING **: Configuring 'ppp-udeb' failed with error code 1 This is the same error Abel faces Hmm, if this is because of ppp-udeb, then it could be because of the apt-install command (shooting in the dark, but its worth a try). Abel, Gregory, could you try to modify (nano is available), before running the PPPoE menu item, the ppp-udeb.postinst script (at the end) It should be in /var/lib/dpkg/ppp-udeb.postinst, sorry for forgetting that info. so it reads: snip # FIXME - how can this be improved? [ -e /etc/resolv.conf ] || ln -s /etc/ppp/resolv.conf /etc/resolv.conf log-output apt-install ppp || true #DEBHELPER# /snip instead of snip # FIXME - how can this be improved? [ -e /etc/resolv.conf ] || ln -s /etc/ppp/resolv.conf /etc/resolv.conf log-output apt-install ppp #DEBHELPER# exit 0 /snip Then tell me if things improve? With logs, if not. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Bug#384709: add a syslog
On 29/08/06, Abel McClendon [EMAIL PROTECTED] wrote: For assistance in debugging a simpler syslog is attached. the failure is towards the end of the log... of course i did the mv patch... going to try again with the latest build. A custom image could be built with the patched ppp-udeb package, as I did when testing my changes in it, and you could test my (yet) unofficial patch. Tell me if you are interested (it can be scripted to the level that you patch the ppp source than run a command which compiles the package, then places the resulted ppp-udeb package in the image, then rebuilds the iso). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#385026: Bug explaination
On 29/08/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: This is it not really a bug in the strict sense of the term, as this is simply the default behaviour of GTK+. Anyway, i'll try to see if it's possible making the enter key activate the ok button (and it should be). What? Are you sure you answered to the right bug? This reports a rendering issue... The text is not displayed when checking the corresponding checkbox. Please see the attached last two pics in the BR. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#385026: Bug explaination
On 29/08/06, Eddy Petrişor [EMAIL PROTECTED] wrote: On 29/08/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: This is it not really a bug in the strict sense of the term, as this is simply the default behaviour of GTK+. Anyway, i'll try to see if it's possible making the enter key activate the ok button (and it should be). What? Are you sure you answered to the right bug? This reports a rendering issue... The text is not displayed when checking the corresponding checkbox. Please see the attached last two pics in the BR. Errr, when selecting a line, not when checking a checkbox. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: Tests with PPPoE connection
clone 384709 -1 reassign -1 cedebconf-gtk-udeb retitle -1 display issues in chekbox lists thanks -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Many comments to Beta 3
On 27/08/06, Joey Hess [EMAIL PROTECTED] wrote: Eddy Petrişor wrote: Indeed, this was discussed many times, but I think the main stopper is their real meaning (which in turn is 0 information for a newbie user). I would suggest: businesscard - keychain netinst remains this way or - minidisc The businesscard CD is targeted at the CD images that are the size and shape of a business card. Due to only having a usable small circular area in the center, these are limited to something less than 100 mb of data. I *think* I have seen only onc such a CD, and I have seen more often minicd-s (about 200MB). Does this hold true for others, too? And netinst is targeted at installation from the net, as opposed to installing from businesscard, for the former? The current naming scheme is inconsistent, IMHO, that's why I think it would be good to improve this. My proposal(with minidisc) would make this possible by allowing to say the name is related to the size of the medium on which the image could be written to. Or, what if we make some ghost images which are in fact symlinks to current buiunesscard images and are named keychain (or usbstick, or whatever is better)? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: Tests with PPPoE connection
clone 384709 -1 # I haven't analysed the whole bug report, but I'll take this ppp-udeb # related issues int consideration for bug -1 reassign -1 ppp-udeb retitle -1 unsanitized output, makes debconf think an error appeared thanks On 27/08/06, Abel McClendon [EMAIL PROTECTED] wrote: check this out and keep the URL under your hat ;-) if someone finds it.. fine... no reason to publicize it.. I am sending a install report that almost mirrors this. Still, the syslog section is not big enough. There should be something Ooops, I've seen now the full syslog link. Sorry. that's OK. here's the filed installation report... http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=384709 Could you try to produce a verbose syslog like this, but which can reporoduce in as few steps as possible the error which appears after trying to up the pppoe connection? (trust ifconfig ;-) That would also mean that you should start a verbose debconf (DEBCONF_DEVL=5), set the debconf priority to low, and just load directly the ppp-udeb component, fix the place where the so file is and run the PPPoE menu entry imediately (without selecting the language and things llike that). I'm afraid that the multiple runs of the ppp-udeb postinst script are a little bit confusing and not having access to a pppoe connection now, makes things even harder. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Tests with PPPoE connection
(this is back in public) On 27/08/06, Abel McClendon [EMAIL PROTECTED] wrote: Things should be easier now, as I have found ou about DEBCONF_DEBUG. Could you please set DEBCONF_DEBUG=5 at boot time and run the PPPoE step (after doing the move of the plugin) and send the relevant log section? this test is _detailed... check this out and keep the URL under your hat ;-) if someone finds it.. fine... no reason to publicize it.. I am sending a install report that almost mirrors this. Still, the syslog section is not big enough. There should be something starting with eth0 was not configured. ppp-udeb will bring it up or eth0 was already up. (even if the relevant Ethernet interface for you is eth1, please make sure you provide also the section with eth0, if any, even if is long; although I don't think so, it could be possible I have messed something in the code and probing is done incompletely). Also, I see your log is starting with: Aug 25 12:55:07 debconf: -- UNREGISTER ppp/password which is part of the current detction, but it is too late, and do not slaughter the log because you are removing precious information (you can obfuscate the addresses, but please, please, please, do NOT remove sections from the syslog). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Tests with PPPoE connection
On 27/08/06, Eddy Petrişor [EMAIL PROTECTED] wrote: (this is back in public) check this out and keep the URL under your hat ;-) if someone finds it.. fine... no reason to publicize it.. I am sending a install report that almost mirrors this. Still, the syslog section is not big enough. There should be something Ooops, I've seen now the full syslog link. Sorry. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: Tests with PPPoE connection
On 23/08/06, Abel McClendon [EMAIL PROTECTED] wrote: hi, actually it's a different error that occurs after the 'mv'. the mv fixes one issue. I understand. the other issue has to do with which physical device is actually connected to the DSL service provider. (usually in my case(s) a bridged modem/router.) I shouldn't matter. The second error is due to somebody spilling messages into debconf. debconf pollution? Now that's not nice ;-) but sounds right. the way the install progresses from netcfg to pppconfig is another one problem. when i do an install over a DSL connection... all i need is the ppp0 for install purposes... I know, it annoyed me also during my tests. It could be fixed by setting a higher priority for the ppp-udeb menu entry (17 instead of 18), which would lead to it being firstly probed. This would make sense since if one loads ppp-udeb it is most likely he/she wants to use ppp and not a regular. The main concern regading this was expressed by Marco (iirc) which said that there are setups where ppp should be ran on an interface which gets its IP via DHCP (allowing the ppp request connection to be whitelisted for that IP). Still I wonder how usual this setup is and I fear that in most cases the right thing would be to just run plain ppp-udeb. d-i won't let me proceed that way(or I haven't found it..) There isn't yet for default install. You need to lower priority and run by hand directly the PPPoE menu entry. PPPoE-Config- -pick the physical device(eth0, eth1, ...(if more than 1) That is picked automatically via probing for a concentrator. Before trying to probe, the interface is up-ed if it was not, otherwise is left up. Interfaces which were up-ed by ppp-udeb, specially for probing are down-ed if probing is unsuccessful. -get username and pass -bring up the provider pppoe(ppp0) -(ppp0) fixes the routing(if not user provide the default) -(ppp0) adds the name servers(if not user provide the name servers) - PPPoE-Config complete That is mainly the way is implemented, except the last part is more ambiguous (saves configuration) and would eb a good thing if it would be closer to what the regular ppp deb does to store and link the configuration (this would allow the settings to play nice with other components and would be the right thing). would work in my case...or something like that... I will retry an install tonight.. (EDT) and attempt to get a much better handle on it. Thanks. Is there a way to not have the netcfg option... just ether detection? That is built in the installer (in the sense that the priority and name of the item determine the order of running them). That priority is 18 for both netcfg and ppp-udeb and, unfortunately, since np, netcfg gets there first. As I said, I think the fix would be to give ppp-udeb a priority lower than 18 (namely 17) so is sorted before netcfg. Detection (in the sense that detects what drivers are needed for the Ethernet interfaces in the system) the will be done automaticaly, since ppp-udeb depends on it. The up-down dance needed to concentrator probing is done correctly by ppp-udeb, AFAI have tested it. In conclusion, if ppp-udeb's prio is lowered to 17, it should do it (unless something in d-i policy forbids it, thing which I don't know/remember to be case). I had the same issue he(Mr. Colpart) had even after i did the patch b4 it asked for user name. I think you may be right the postinst script is picking a non-fatal error. I *know* I am right ;-) , I have seen this symptom one too many times to not understand what is the cause ;-) -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Tests with PPPoE connection
On 24/08/06, Gregory Colpart [EMAIL PROTECTED] wrote: On Wed, Aug 23, 2006 at 03:58:07PM +0300, Eddy Petri??or wrote: Then my PPPoE connection is up correctly (I can wget web files on busybox) even I have the red screen [!!] Configure and start a PPPoE connection, Installation step failed: An installation step failed. [...] The failing step is: multiselect backup progresscancel plugin-entropy-text. after entering PPP password. Hmm, you are saying that you get this error even if you move the pppoe.so plugin in the proper place? Yes. Well, I'll have to see which is the prick, but I will have to lookover the postinst code to see what is changed from what I sent to what has actually been released (I went great lengths to make sure nothing is spilled into debconf and wasted a lot of time on this, just because this was annoying and frustrating, but this resulted in some ugly looking code, which was modified by Marco at merge time). Things should be easier now, as I have found ou about DEBCONF_DEBUG. Could you please set DEBCONF_DEBUG=5 at boot time and run the PPPoE step (after doing the move of the plugin) and send the relevant log section? Relevant means from where it says the PPPoE manu was selected until the place where it gives the errorwith multiselect backup. You can save the logs via the menu entry: http://d-i.alioth.debian.org/manual/en.i386/ch06s03.html#save-logs That error is the classical case of debconf polution, so I suspect some unexpected output from either of the commands in the postinst script. May I suggest doing the move from console 2 before entering the PPP username. Please tell me if the error message still appears. I did the move from tty2 before entering in PPPoE menu. Then error message still appears. I need that output I was talking before. Maybe is time to compare the patch I sent in BTS with the one that ended up in the official package. When I select 'continue', I return on 'Configure the network' entry but my network connection is now ready. I will try to retry/investiguate tomorrow. So the PPPoE connection is actually up? Yes, and works fine. That means that there is a message which is printed by some command which is not redirected to /dev/null and ends up as input for debconf. Note that (almost?) always in my tests the concentrator was not found from the first attempt after a fresh boot, but the search did succeed in later probes. Have you ever encountered this problem? (I was never sure if this was a provider issue or a coding issue and I would be glad to sort it out.) With my two or three tests, concentrator was found from the first attempt after fresh boot. Cool, then it could have been my test machine or my old ISP (which is not at all implausible). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Many comments to Beta 3
On 24/08/06, Jens Seidel [EMAIL PROTECTED] wrote: Hi, (please CC: me) I tested the current Beta 3 of the installer (my aim was not to install a new system, that's why I send no installation report) and want to inform you about my experiences. There are many little issues which are mostly not very important. I used the netinst image from http://cdimage.debian.org/cdimage/etch_di_beta3/i386/iso-cd/. The first confusing fact are the names: debian-testing-i386-businesscard.iso (38MB) debian-testing-i386-netinst.iso (139MB) I know this has been already discussed in the past but now I have a local pocket CD available and noticed that it fits up to 210 MB. So it's sufficient large for a netinst CD. The businesscard ISO name is that's why confusing. Why not swap the names? (A netinst CD is also useful as rescue system, ... without network connection at all, a businesscard CD probably not.) Indeed, this was discussed many times, but I think the main stopper is their real meaning (which in turn is 0 information for a newbie user). I would suggest: businesscard - keychain netinst remains this way or - minidisc The following refers to the graphical user frontend and expert mode but I also tested the newt and partially text frontend: There is a Screenshot button which I cannot select/use. I tried TAB and cursor keys, no success! An Alt+S shortcut (which is visible via: _Screenshot) would be nice. I tried also my print key but the button appearance did not change (no pressed+release animation or something fancy like this). I think I have seen at some point the request to make screeshots via the Print key possible, or, even better via Alt+S. Probably, if a bug is not filled (I have looked and haven't found one with an obvious name), it should be filled. I have also an ugly mouse cursor in the middle of the screen. Since the mouse does not work I suggest to hide it (e.g. by moving it to the bottom right corner). Maybe, placing it there, but over the screenshot button :) . Seriousely, it should be visible, but in a no-button area, so users with a mouse don't get confused. Or maybe is possible to detect non-working mouses and hide the cursor? The string Choose language at the beginning of the installation process (e.g. during the locale selction) is not translated even after I selected German. After a Go Back and reconfiguration it changed into the mixed term Sprache wählen/Choose language. The language setting is enabled after is selected and localchooser runs. It is not selected on the spot. Also untranslated: Configure and start a PPPoE connection (main menu) For that a translation for the po-debconf template of the ppp package is needed, which is not a part of D-I nor its levels (yet). After ppp-udeb gets more testing it might become. But that is no reason to not send a new or an updated translation into BTS. I get the message This computer may have a PCMCIA interface. but I definitively have no PCMCIA interface, only two ordinary ethernet cards (one is a ISA card). It says may have, it lets you decide that ;-) . AFAIK there is no reliable way to say yes or no ATM. My PC is connected via eth0 with my notebook which itself is connected (via eth2) to my router. There is no direct connection to the router. I enter my notebook as gateway but cannot specify a nameserver (I have not installed bind). I would like to use my router as nameserver but this requires a route add -host router ip gw notebook ip which I currently start manually from another console or ssh. I know that there exists various solutions (nevertheless I like my current configuration): * installation of a nameserver on the notebook * direct connection to the router * DHCP Why not configure your DHCP server to provide the correct nameserver and router options? May I suggest dnsmasq, which is really good at this thing. I was also once able to enter the Continue with SSH menu without a current network configuration (the IP adress is the dialog was empty, also the ifconfig output). IIRC I did the following: I entered a proper static network and selected later PPPoE which failed and destroyed my configuration. Go Back in Detect disks goes to partitioner, not to main menu. That is because you are inside partman. That is the correct behaviour, and is documented in the manual. I cannot revert a proposed LVM partitioning of a whole disk. Nothing happens if I select revert. I can remove all physical volumes manually but the related logical volumes (which use removed PV!) cannot be removed! AFAIK, you have to configure LVM accordingly and delete those first. Another issue: I found a reference to the netinst CD four times in my /etc/apt/sources! Two are commented out. I think this has been reported (and fixed?) recently. Once for each time apt sources menu was visited. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: Bug#384121: ppp-udeb: fails to find rp-pppoe.so
On 24/08/06, Eddy Petrişor [EMAIL PROTECTED] wrote:
Hello Marco,
(Pathes were intentionaly not forwarded since they are already in BTS)
I just realised I sent this to -quiet.
A fix for this bug is attached. Please note that I have updated the
Romanian translation, too.
Now I have a patch which should fixes the old issue and, in addition,
places the PPPoE menu entry before the netcfg menu entry, thus
allowing a configuratio for a PPPoE only install system, which seems
to be more frequent [1].
Although Abel and Gregory are willing to test the ppp-udeb now, I
think it would be a good thing to update the package with this fix,
and why not my translation because this bug is slowing down their
testing efforts, and fixing it would help them a lot.
This update should also make it more easier for them to test.
[1] http://lists.debian.org/debian-boot/2006/08/msg01124.html
--
Regards,
EddyP
=
Imagination is more important than knowledge A.Einstein
diff -ruN ppp-2.4.4rel.orig/debian/changelog ppp-2.4.4rel/debian/changelog
--- ppp-2.4.4rel.orig/debian/changelog 2006-08-24 22:43:54.0 +0300
+++ ppp-2.4.4rel/debian/changelog 2006-08-24 22:50:18.0 +0300
@@ -1,3 +1,13 @@
+ppp (2.4.4rel-3) unstable; urgency=low
+
+ [ Eddy PetriÈor ]
+ * fix the place where rp-pppoe.so is placed in (Closes: #384121)
+ * to ppp-udeb a higher priority in the menu so it runs before netcfg since
+loading ppp-udeb means, most likely, the user wants to install via PPPoE
+and that would be the primary connection type.
+
+ -- Eddy PetriÈor [EMAIL PROTECTED] Thu, 24 Aug 2006 22:47:18 +0300
+
ppp (2.4.4rel-2) unstable; urgency=high
* Make postinst not fail if /dev/MAKEDEV does not exist yet.
@@ -5,7 +15,7 @@
* Update debconf translation: da. (Closes: #377672)
-- Marco d'Itri [EMAIL PROTECTED] Wed, 12 Jul 2006 18:26:01 +0200
-
+
ppp (2.4.4rel-1) unstable; urgency=high
* New upstream release. Fixes:
diff -ruN ppp-2.4.4rel.orig/debian/control ppp-2.4.4rel/debian/control
--- ppp-2.4.4rel.orig/debian/control2006-08-24 22:43:54.0 +0300
+++ ppp-2.4.4rel/debian/control 2006-08-24 22:47:04.0 +0300
@@ -25,7 +25,7 @@
Architecture: any
Depends: ${shlibs:Depends}, ethernet-card-detection, ppp-modules, di-utils
Provides: configured-network
-XB-Installer-Menu-Item: 18
+XB-Installer-Menu-Item: 17
Description: Point-to-Point Protocol (PPP) daemon
The Point-to-Point Protocol (PPP) provides a standard way to transmit
datagrams over a serial link, as well as a standard way for the machines
diff -ruN ppp-2.4.4rel.orig/debian/patches/000_fix_internal_version.diff
ppp-2.4.4rel/debian/patches/000_fix_internal_version.diff
--- ppp-2.4.4rel.orig/debian/patches/000_fix_internal_version.diff
1970-01-01 02:00:00.0 +0200
+++ ppp-2.4.4rel/debian/patches/000_fix_internal_version.diff 2006-08-24
22:46:30.0 +0300
@@ -0,0 +1,10 @@
+Binary files ppp-2.4.4.orig/.ilist and ppp-2.4.4/.ilist differ
+diff -ruN ppp-2.4.4.orig/pppd/patchlevel.h ppp-2.4.4/pppd/patchlevel.h
+--- ppp-2.4.4.orig/pppd/patchlevel.h 2006-06-28 03:21:23.0 +0300
ppp-2.4.4/pppd/patchlevel.h2006-08-24 17:36:00.0 +0300
+@@ -1,4 +1,4 @@
+ /* $Id: patchlevel.h,v 1.66 2006/06/28 00:21:23 paulus Exp $ */
+
+-#define VERSION 2.4.4
++#define VERSION 2.4.4b1
+ #define DATE 28 June 2006
Binary files ppp-2.4.4rel.orig/.ilist and ppp-2.4.4rel/.ilist differ
Re: Tests with PPPoE connection
On 23/08/06, Gregory Colpart [EMAIL PROTECTED] wrote: On Tue, Aug 22, 2006 at 10:10:08PM -0400, Abel McClendon wrote: check bug #384121 Indeed, it's exactly same problem. CTL-alt-F2 and issue the command mv /usr/lib/pppd/2.4.4b1 /usr/lib/pppd/2.4.4 With this workaround, I retry my tests with PPPoE connection. Then my PPPoE connection is up correctly (I can wget web files on busybox) even I have the red screen [!!] Configure and start a PPPoE connection, Installation step failed: An installation step failed. [...] The failing step is: multiselect backup progresscancel plugin-entropy-text. after entering PPP password. Hmm, you are saying that you get this error even if you move the pppoe.so plugin in the proper place? That error is the classical case of debconf polution, so I suspect some unexpected output from either of the commands in the postinst script. May I suggest doing the move from console 2 before entering the PPP username. Please tell me if the error message still appears. Maybe is time to compare the patch I sent in BTS with the one that ended up in the official package. When I select 'continue', I return on 'Configure the network' entry but my network connection is now ready. I will try to retry/investiguate tomorrow. So the PPPoE connection is actually up? Note that (almost?) always in my tests the concentrator was not found from the first attempt after a fresh boot, but the search did succeed in later probes. Have you ever encountered this problem? (I was never sure if this was a provider issue or a coding issue and I would be glad to sort it out.) Thanks for testing this. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: r40000 - tags/packages/installation-report/2.18
On 20/08/06, Frans Pop [EMAIL PROTECTED] wrote: On Sunday 20 August 2006 01:13, Frans Pop wrote: Author: fjp Date: Sat Aug 19 23:13:50 2006 New Revision: 4 ^ Another milestone reached... I still wonder why we don't consider milestones at revisions with numers representing 2's powers :) For comparison: Rev 35000 - bubulle - 2006-02-19 13:30:41 + (Sun, 19 Feb 2006) Rev 3 - joeyh- 2005-08-17 16:28:20 + (Wed, 17 Aug 2005) Rev 25000 - cjwatson - 2005-01-17 10:54:55 + (Mon, 17 Jan 2005) Rev 2 - smarenka - 2004-08-26 02:26:39 + (Thu, 26 Aug 2004) Rev 15000 - joeyh- 2004-05-08 23:52:49 + (Sat, 08 May 2004) Rev 11500~ - 2004-03-21 - d-i moved from cvs to svn Rev 1 - zinoviev - 2004-03-06 19:42:01 + (Sat, 06 Mar 2004) Rev 5000 - goswin - 2003-10-16 23:09:10 + (Thu, 16 Oct 2003) Rev 1 - joeyh- 2000-06-21 17:47:53 + (Wed, 21 Jun 2000) The last one has the following commit message: Moved in from my own cvs repository. Seems we are fairly steady at about 50.000 commits per half year since I think you mean 5000/half a year ;-) 2004, although there are obvious peaks. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Why manual po-file msgids one long line?
On 18/08/06, Tapio Lehtonen [EMAIL PROTECTED] wrote: Is it necessary to have the installer guide msgids in po files one huge line? It would seem to me they could be indented to make them more readable. I mean, text like this is pretty bad for reading: msgid The benefit you gain depends on a type of a MD device you are creating. Currently supported are: variablelist varlistentry termRAID0/ termlistitempara Is mainly aimed at performance. RAID0 splits all incoming data into firsttermstripes/firstterm and distributes them equally over each disk in the array. This can increase the speed of read/ write operations, but when one of the disks fails, you will loose emphasiseverything/emphasis (part of the information is still on the healthy disk(s), the other part emphasiswas/emphasis on the failed disk). /parapara The typical use for RAID0 is a partition for video editing. /para/listitem /varlistentry varlistentry termRAID1/ termlistitempara Is suitable for setups where reliability is the first concern. It consists of several (usually two) equally sized partitions where every partition contains exactly the same data. This essentially means three things. First, if one of your disks fails, you still have the data mirrored on the remaining disks. Second, you can use only a fraction of the available capacity (more precisely, it is the size of the smallest partition in the RAID). Third, file reads are load balanced among the disks, which can improve performance on a server, such as a file server, that tends to be ... Judging from the title I would have said that you wanted the layout above :-/ msgcat can do this. I did not find a readymade program for indenting xml, but I could write some python code to make the above indented sensibly. I was planning setting a max line length and starting a new line for tags like listitem, para, varlistentry etc. Probably that would be a plus, but don't forget to keep spaces between words where needed. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Why manual po-file msgids one long line?
On 18/08/06, Tapio Lehtonen [EMAIL PROTECTED] wrote: Is it necessary to have the installer guide msgids in po files one huge line? It would seem to me they could be indented to make them more readable. I mean, text like this is pretty bad for reading: msgid The benefit you gain depends on a type of a MD device you are creating. Currently supported are: variablelist varlistentry termRAID0/ termlistitempara Is mainly aimed at performance. RAID0 splits all incoming data into firsttermstripes/firstterm and distributes them equally over each disk in the array. This can increase the speed of read/ write operations, but when one of the disks fails, you will loose emphasiseverything/emphasis (part of the information is still on the healthy disk(s), the other part emphasiswas/emphasis on the failed disk). /parapara The typical use for RAID0 is a partition for video editing. /para/listitem /varlistentry varlistentry termRAID1/ termlistitempara Is suitable for setups where reliability is the first concern. It consists of several (usually two) equally sized partitions where every partition contains exactly the same data. This essentially means three things. First, if one of your disks fails, you still have the data mirrored on the remaining disks. Second, you can use only a fraction of the available capacity (more precisely, it is the size of the smallest partition in the RAID). Third, file reads are load balanced among the disks, which can improve performance on a server, such as a file server, that tends to be ... Judging from the title I would have said that you wanted the layout above :-/ msgcat can do this. I did not find a readymade program for indenting xml, but I could write some python code to make the above indented sensibly. I was planning setting a max line length and starting a new line for tags like listitem, para, varlistentry etc. Probably that would be a plus, but don't forget to keep spaces between words where needed. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A couple of small udebs
On 17/08/06, Stephen Frazier [EMAIL PROTECTED] wrote: I am interested in using the first two udebs (cfg-netcfg and cfg-partman). How do I get a copy of them? How do I set up ETCH to use them? If you want the binaries you can grab them from the mirrors (under debian-installer, instead of debian, as you would use on a regular system), but please note that THESE ARE NOT TO BE USED IN A REAL SYSTEM. They are specially crafted for D-I environment. If you want the sources, you can grab them from svn. svn://svn.debian.org/svn/d-i/trunk/packages/ (browsable via web http://svn.debian.org/wsvn/d-i/trunk/packages/) -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: General question
On 02/08/06, Alexander Schmehl [EMAIL PROTECTED] wrote: Note that, generally the defaults chosen in the installer are prety decent and in most cases you will need to adjust questions' priority minimal level that needs to be shown. Actually I was planing to create such a Debian Desktop image since CeBIT (we created a nice preceeding DVD there), and later, if it has proven stable and usefull, come back to you guys. Sadly I never had time to do it :( That's why automated partitioning exists (question for others, is automated paritioning in free space an option currently present in partman?). Hmmm... I have checked it yesterday, it does not. This would be a really nice feature, although I know partman is no longer maintained actively... -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: General question
On 02/08/06, Frans Pop [EMAIL PROTECTED] wrote: On Wednesday 02 August 2006 13:18, Eddy Petrişor wrote: That's why automated partitioning exists (question for others, is automated paritioning in free space an option currently present in partman?). I have checked it yesterday, it does not. This would be a really nice feature, although I know partman is no longer maintained actively... I think you are mistaken. There still is an option to automatically partition using free space on a harddisk. Does that only show up when free space is available? That would explain why I haven't seen it when I tried the yesterday's daily. Also, I fixed a huge load of bugs in partman recently, we gained support for crypto and LVM had a major rewrite (thanks to Max and David). So partman is not totally unmaintained either. I know about those, but I said actively maintained and was referring also the the rest of the partman code base, Anton's work ;-) , which I think is fair to say it is not actively mainainted (which can be seen quite clearly, IMHO, when looking at the age of partman's bugs). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: General question
On 31/07/06, Hans-J. Ullrich [EMAIL PROTECTED] wrote: Dear developers, Hello Hans I would be pleased, if you could tell me your thoughts about some points, which came in my mind. 1. In the past I discovered the project FAI by Thomas Lange and changed it to the possibility to install not as Thomas` version Debian sarge, but Debian etch (and sid,too). Everything worked just fine. So my idea was, to implement this famous tool into the installer as an option, to get a fast installation for newcomers. I imagine it as a standard installation, where the newbie will get a ready system, which he only has to fine tune. What do you think, should this kind of option be integrated in the future ? There is preseeding which should automate installations for as many cases as you may want. But, OTOH, I think it would be useful to add some means to select at boot time one of newbie scheme, mail server scheme, desktop scheme ... etc. This way, it would be really easy to put on the splash screen soemthing like Type fast install for a fast and simple desktop installation - which should install a desktop newbie system. Note that, generally the defaults chosen in the installer are prety decent and in most cases you will need to adjust questions' priority minimal level that needs to be shown. 2. Tasksel: To use of tasksel is very fine I think. But what I really miss is the option to choose different profiles like: Desktop, Webserver, Mailserver, Development etc. I saw an example on the Debian ct` server CD, which was Hmm, such tasks exist, at least for desktop, laptop and file server, iirc. Please understand my intention: To install Debian is a peace of cake for me, but for a very newcomer it should be as easy as it could be. A newcomer even does not know, what is a partition (primary ? logical ) , he does not know anything about IP or MBR. That's why automated partitioning exists (question for others, is automated paritioning in free space an option currently present in partman?). Also THE MANUAL has very good explanations about such things AND it can be browsed on a jump to interesting item fashion. The knowledge will appear later, when he has used Debian for a while. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#379120: lspci present on i386, verify on powerpc
On 29/07/06, Geert Stappers [EMAIL PROTECTED] wrote: On a i386 is lspci present in the recent builds op debian-installer. While the installer waits for a keyboard selection, I switched to the second console and typed `lspci` and got the output of lspci. Could this be verified on a powerpc computer? (no need to do a re-installed ( no computer harm )) The gtk-miniiso (downloaded a week ago, IIRC) contained it, just cjhecked. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#380093: Debian branding
On 30/07/06, Colin Watson [EMAIL PROTECTED] wrote: On Fri, Jul 28, 2006 at 02:49:53PM +0300, Eddy Petrisor wrote: Colin Watson wrote: It would be nice if this text were more generic so that derived distributions didn't have to mess with it. How about only the core of your new system is installed and Software selection? May I suggest removing your from your proposed text? It seems more professional to me. Simply removing your from the text above would produce an ungrammatical sentence. I disagree anyway - your is used all over the Indeed, I should have said replace it with 'the' , but Joey got it right IMHO :) -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#380093: Debian branding
On 28/07/06, Joey Hess [EMAIL PROTECTED] wrote: Eddy Petrisor wrote: May I suggest removing your from your proposed text? It seems more professional to me. At the moment, only the core of the system is installed. To tune the system to your needs, you can choose to install one or more of the following predefined collections of software. I think I removed the first your already because it scanned better. Indeed, this way seems better (actually, that exact your bothered me, but I might have not been too clear on that). Also, note that in some situations the installation is not done on a system owned by the person doing the installation, so is more correct *not* to say your system ;-) . -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: d-i with GRUB
On 28/07/06, Robert Millan [EMAIL PROTECTED] wrote: Hi! Hello! I've made a proof-of-concept d-i build with grub as the CD loader: http://aybabtu.com/rmh/d-i_with_grub.iso cool Pros: - boot screen not shown at all unless requested with ESC (avoids confusing newbie users). - options are presented in a menu with long, self-descriptive option titles - any of the boot params can be modified with user-friendly line editor - additional option, Boot from harddisk, that chainloads into MBR (currently not possible). - access to GRUB console by pressing c, which enables advanced users to browse filesystems, probe partitions, etc. Cons: - no pretty splash image (this could be compensated with Linux bootsplash support, though). Why? Grub supports splash images. Any reason not to use that (except the size of the initrd?) - no context function-key help. You could fix that with a nice splash image ;-) What do you think? Would you like a patch? Probably is good to have that in BTS, even if is postponed until etch+1, so I would say yes. Later updates you might do can be sent to the existing bug. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#379521: marked as done (Install report, focusing on Hebrew l10n issues)
On 24/07/06, Debian Bug Tracking System [EMAIL PROTECTED] wrote: The installer started in text mode and not g-i, I assume it's something to do with qemu. Note that the installer does no0t default yet to GUI, you need to boot with installgui to start the graphical installer. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Installing Debian unstable to IBM pSeries LPAR - success
Please submit an bug against the installation-reports meta package. (Hint: using reportbug will allow to generate a proper installation report). On 24/07/06, AlexB [EMAIL PROTECTED] wrote: Good day everyone, just managed to install Debian unstable to an IBM p570 LPAR. Generally installation went OK, although Debian installer complained about inability to finish installation of uw-imapd at the end; also installer complained about inability to install linux kernel properly, but kernel was installed. Now, after restart, several small issues have been spotted: 1) eth0 interface was not brought up, seems to be a typo in /etc/network/interfaces, 'auto eth0' line was missing. 2) no entry for console getty in /etc/inittab - so no way to log in from console (the console is a terminal connection to virtual seriel port from HMC) 3) no openssh server installed by default - so even if ethernet is brought up, there is no way to log in to the system after reboot due to point 2). After openssh server installation - no startup script in /etc/rcS.d, the link has to be done manually. The only way to correct was to boot with 'single' kernel parameter and modify relevant files by hand. After reboot everything seems to be OK. /etc/issue reads: Debian GNU/Linux testing/unstable \n \l uname -a reads: Linux lnxsrv 2.6.15-1-powerpc64 #2 SMP Mon Mar 6 12:50:57 CET 2006 ppc64 GNU/Linux I used debian-testing-powerpc-netinst.iso image as of 17/07/2006 and installed rest of packaged over Internet. Alex -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Boot problem after flasing, led is not green.
On 24/07/06, Joey Hess [EMAIL PROTECTED] wrote: Jean-Pierre Broeders wrote: I have a problem after i flashed for the last time using this manual: http://www.cyrius.com/debian/nslu2/ After is flashed debian-sda1.bin, I shutdown the power and connect my usb harddisk. When I power my nslu2 the only green led (also after some couple of hours) i see is the transfer led. The ready/status is orange. I tried to log in with SSH, but it also doesnt work. What can be wrong? Or am i forgotten something? An amber status led and green network led is actually normal for a running nslu2 system with debian on it right now (just checked mine). We haven't sorted out the necessary stuff to turn the status and disk leds green. ssh not working of course indicates a problem. Can you ping the system? I have encountered a similar problem about a half year ago, talked with Martin Michmayr about it, but we couldn't spot the reason why this happens. I eneded up using debian armeb unofficial port, but solving this issue would be a real plus. Note: I find it very useful that the status led blinks when the NSLU is powered and running in debian armeb since I have experience seldom blockage of the router and at those times the blinking stops and is easy to spot. Still I am not sure if this is a side effect or the intended behaviour, but is nice. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Hint for iso-codes
On 21/07/06, Tobias Toedter [EMAIL PROTECTED] wrote: Hi, the package iso-codes could be hinted into testing today, if the D-I team doesn't have any objections. So RO is not 100% in the beta release :( #378665 (judging from the changelog) ? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#370426: libgtk+2.0-directfb-dev: please update libdirectfb dependency
On 21/07/06, Laurent Bonnaud [EMAIL PROTECTED] wrote: Hi, this bug is fixed in experimental. Thanks! What prevents you to upload the updated version into unstable? The fact that the GNOME team will use which ever gtk+2.0 version they see fit in order not to break GNOME. Until the changes in gtk+2.0 version 2.8.18-3 do not reach unstable, we can't fix it. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#379067: couple of minor glitches in install of powerpc
On 21/07/06, Rick Thomas [EMAIL PROTECTED] wrote: On Jul 21, 2006, at 4:37 AM, Wouter Verhelst wrote: On Thu, Jul 20, 2006 at 09:13:51PM -0400, Rick Thomas wrote: Some testing also noted that /etc/apt/sources.list includes the install CD as a source. I don't think this is a good idea -- it means that I have to hang onto the install CD and put it in the drive every time I want to apt-get something. That's just silly. No, it isn't. If you installed from CD-ROM, that may be because you bought a CD-ROM from some CD reseller because your internet connection is paid for by the minute and you don't want to have to download too much. Or you may not have an internet connection at all. If you don't want the CD-ROM as an install source, it's pretty easy to edit sources.list and throw out the CD-ROM URI. Sorry. I wasn't clear. I'm installing from the businesscard installer CD. I call BS :-D ! Joke aside, this i from a system installed on Monday via a buisnesscard CD: [EMAIL PROTECTED]:~/var/cache/chroots/rpm/stentz-rpm$ cat /etc/apt/sources.list # deb http://ftp.ro.debian.org/debian/ etch main # deb http://ftp.ro.debian.org/debian/ etch main # Line commented out by installer because it failed to verify: # deb-src http://ftp.ro.debian.org/debian/ etch main # # # deb http://security.debian.org/ etch/updates main # deb-src http://security.debian.org/ etch/updates main # deb http://ftp.ro.debian.org/debian/ etch main non-free contrib # deb-src http://ftp.ro.debian.org/debian/ etch main non-free contrib # # # deb http://security.debian.org/ etch/updates main contrib non-free # deb-src http://security.debian.org/ etch/updates main contrib non-free deb http://ftp.ro.debian.org/debian/ etch main non-free contrib deb-src http://ftp.ro.debian.org/debian/ etch main non-free contrib deb http://security.debian.org/ etch/updates main contrib non-free deb-src http://security.debian.org/ etch/updates main contrib non-free I don't see any CD source. Maybe you were talking about netinst? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#378971: package: installation-reports
On 20/07/06, Mark Pictor [EMAIL PROTECTED] wrote: Something else: extended partitions aren't supposed to be bootable, are they? Yet sda7 was set to boot, and it is extended, not primary. sounds like a bug to me... You are confusing logical partitions with extended partitions. The later type can contain the former type. Device Boot Start End Blocks Id System /dev/sda1 1 5 40131 83 Linux /dev/sda2 6 12163976591357 HPFS/NTFS /dev/sda3 12164 36481 1953343355 Extended ^^^ this is the only extended partition (the extended partition is also primary) \/\/\/\/\/\/\/\/\/\/\/\/\/\/\/ these are all secondary paritions /dev/sda5 12164 2414196213253+ c W95 FAT32 (LBA) /dev/sda6 24142 3629997659103+ 83 Linux /dev/sda7 36300 36481 1461883+ 83 Linux sda1 - /boot for Etch AMD64 sda2 - C: for WXP SP2 sda5 - D:, FAT32 so it can be written from WXP and Linux sda6 - / for Etch AMD64 sda7 - / for Etch x86 (this is the install I have trouble with) In your case I suppose the partition which should be bootable is sda2 -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#378706: Acknowledgement (installation-reports: succesful installation on Dell Optiplex GX520)
On 18/07/06, Debian Bug Tracking System [EMAIL PROTECTED] wrote: Of course, the installation was done yesterday with an image built on 20060716 (July not June). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: ppp-udeb debconf translations and D-I levels
On 15/07/06, Frans Pop [EMAIL PROTECTED] wrote: On Saturday 15 July 2006 09:40, Christian Perrier wrote: Recent ppp package versions have bringed an enhanced ppp-udeb package which allows configuring a PPPoE connection for the installer. Thanks a lot for this work. I have not really seen any test results for the latest version of the package. How does it work now? Is the system also correctly set up to use PPPoE after the reboot? The system, in my last tests before my vacation, was not left in a correct state although I was confident they should have. The installation was done over pppoe correctly on my setup. I am not sure I will have the time (in the near future) to investigate why the system was not set properly, becauseI will move to Bucharest soon and I am not sure when I will have everything set in place again to make it possible to work at home. This definitely seems to be a very interesting feature. I don't know whether Frans will be OK to include it as a more standard component of D-I (currently, being optional, it has to be explicitely added when using D-I at medium priority) but this seems quite likely to happen at some moment. I'm not sure. Including it by default also means pulling in the dependencies (total installed size ~0.5 MB). More importantly, running the menu item included in the udeb [1] currently generates a huge error if no concentrators are found and leaves my network configuration completely broken afterwards (running netcfg again does not fix it). The network configuration, in case the concentrator is not found, should not be affected. It is true that currently ppp-udeb does not provide configured-network, although it should. It is strange, you say that running netcfg does not fix the thing, it should leave the thing untouched. It seems to me that the changes also make the new version of the udeb less suitable for installations using a real modem. However, I'm not sure as I've never tried an installation using a real modem. Using a real modem was never, AFAIK, a target for the udeb. I feel that making the ppp-udeb detect and work correctly for both pppoe and real modem would need a ppp-connection-detector-udeb. The logic inside the postinst of ppp-udeb is already too hairy. From the reports we have received over time PPPoE is only needed by a relatively small number of users. A better option is probably to properly document things in the installation guide and the FAQ. I was wondering, can a module loading be commanded via a boot parameter? That would simplify the ppp-udeb loading sequence a lot. The udeb probably could be improved by first asking if the user is using a real modem or wants to use PPPoE and then proceeding accordingly. Probably, but plain modem functionality is not implemented at all in the postinst, ATM. If the udeb is now targeted primairily at PPPoE installations, Marco should probably change its short description as the udeb is still listed as: Point-to-point Protocol (PPP) daemon By hand a ppp connection could be set. That configuration should be copied in the target, as the code is written now (and as it should work ;-) But you are right, is not doing exactly that, setting automatically any PPP connection, just PPPoE ones. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: ppp-udeb debconf translations and D-I levels
On 16/07/06, Marco d'Itri [EMAIL PROTECTED] wrote: [EMAIL PROTECTED] wrote: ADSL does not imply PPPoE though. I have ADSL too, but don't need PPPoE. In some countries (e.g. DE, IT) is always does for consumer connections. Add RO to that list :-) -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Please fix translation errors
On 16/07/06, Davide Viti [EMAIL PROTECTED] wrote: ATM there are 30+55 errors (suspect variables + level specific errors) in level1 translations (see [1]). Languages currently affected: [snip] Please spend a few minutes to fix those so that we'll have a (nearly) error-free set of translations for next release:) Please do not hesitate to ask if there are any doubts. Maybe is a good idea to say that translators should understand that the kind of errors Davide is talking about, if not fixed, can lead, in some cases, to a non-functional Debian-Installer in the language for which errors exist. This would make their efforts useless. In cocnlusion, please fix those errors! -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
[VAC] 7th of July - 17th of July
Hello all, Please excuse the spam, just wanted to announce I will not be available during the next week as I will be on vacation. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GTK+ 2.8.18 experimental debs/udebs
On 04/07/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: Where from can I get the gtk-demo app source? usually it's built togheter with gtk libraries : you should have it already compiled and ready to run somewhere in the gtkdfb packages you built.. But those are linked against the X11 variant, will have o rebuild it. Unfortunately, I don't think I will have time for this as I will be on vacation starting tomorrow. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#279855: ppp-udeb: pppoe support in D-I
Package: ppp-udeb Severity: wishlist Tags: patch, d-i, l10n Hello, This patch should provide pppoe functionality to the D-I images, if loaded and ran from D-I itself. It also provides a Romanian translation of the used debconf templates. It still needs some polishing and some answers[1][2], but is generally functional. The most important things which need to be fixed are: - make finish-install.d code work correctly so pppoe configuration is preserved - see why concentrator detection does not work imediately after a reboot and fix it - make the postinst script idempotent All these points and other minor ones are (or at least I remember I have put in the debian/TODO.ppp-udeb file ;-) [1] http://lists.debian.org/debian-boot/2006/05/msg00440.html [2] see the message below, which is not yet visible in the web interface, ATM -- Forwarded message -- From: Eddy Petrisor [EMAIL PROTECTED] Date: 05-Jul-2006 09:04 Subject: ppp-udeb: I have managed to install the system via pppoe To: debian-boot@lists.debian.org -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Note: I had this mail in my drafts directory for some time now; as I see I don't have time to fix it due to real life changes and issues, I will send it now, before fixing the last bits. Hello all, The modifications I done on the ppp-udeb package, in order to make it really work in the D-I environment, are functional. I have made a modified iso image[1], so beta testers will only need to download that and use it. It still needs some polishing, but the main issue, the default route which was not set is a weird issue since, although the route command displays the default route 0.0.0.0, the installation went fine via pppoe (it uses finish-install.d). The not-so-important things to do now are: - - use the same scheme for additions to /etc/resolv.conf as the general pppd package uses; (this also means having a separate file for the pppoe peer configuration, and not having that in the postinst) - - make the postinst really reentrant by detecting any previous instances of pppd and killing them before rerunning - - do not return to the main menu until actually a pppX interface is up and has an IP, or until a time out (gah, timeouts) - - see if concentrator detection really works for anybody from the first attempt (I even copied pppoeconf code to try to fix this, but I didn't had luck) I still couldn't figure out why the concentrator is never detected at the first attempt, meaning that in every test I attempted immediately after a boot, the concentrator was not found. I tried to run pppoe-discovery twice with different parameters in the hope that it will work (as seen in pppoeconf). The conclusion is that I think this is material for beta testing (NOT in the sense D-I beta testing, but in the sense that someone else should try it, test it). The code will need a review, I am sure I could have done some things a lot more cleaner, than they are now. I also feel that the code should be integrated in the official package because this iteration is too big already. I will probably make time for the rest of the points (except for the concentrator detection issue, where I am requesting help) P.S.: Just discovered that the finish-install.d thingie does not work :-/ [1] http://eddyp.homelinux.net:8080/eddy/pppoe/current/pppoe-di-i386.iso - -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.3 (GNU/Linux) Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org iD8DBQFEq1aKY8Chqv3NRNoRApHOAJ9I2MpFxSX08poxwWT4fmvKTm+BAgCcC/12 E1CsU4gSj2IgMoPOZvX4Y50= =CBsQ -END PGP SIGNATURE- -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#279855: ppp-udeb: pppoe support in D-I
Forgot the patch On 05/07/06, Eddy Petrişor [EMAIL PROTECTED] wrote: Package: ppp-udeb Severity: wishlist Tags: patch, d-i, l10n Hello, This patch should provide pppoe functionality to the D-I images, if -- Regards, EddyP = Imagination is more important than knowledge A.Einstein ppp-udeb_templates_and_logging.patch.gz Description: GNU Zip compressed data
Re: A web-based frontend for the cdebconf
On 04/07/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: This evening i built a test tarball, similar to those i used to build when developing the GTK frontend, to test the web frontend with thttpd in a real d-i environement. The results were encouraging, even if a lot of work has to be done yet. There are many bugs in the code still awaiting to be fixed and major work is needed to support HTTP encoding/decoding and to correctly display non latin languages [1]. Ouch, I think this is going to be a major problem. IIRC, a web page can have only one encoding, thus we will hit again the Han unification problem (Thanks a lot Unicode consortium for performing a lobotomy on the unicode!). If anybody can contradict me, please, do. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Plans for GTK+ - update
On 03/07/06, Frans Pop [EMAIL PROTECTED] wrote: On Monday 03 July 2006 11:51, Josselin Mouette wrote: Le vendredi 30 juin 2006 à 19:21 +0200, Josselin Mouette a écrit : In the next days, the stable 1.2 branch of libcairo should be released. This has just happened. It has been uploaded to unstable today. Hmm. I'm not totally happy that this happened without coordinating it first. Luckily it will not break the current graphical installer (as the udeb was named differently by Dave that the current hacked one). I think Josselin was reffering to upstream. Sometime during summer, GTK+ 2.10 should be released, with integrated DirectFB support. This has also happened. At one point you feel like you're waiting endlessly for these releases to happen, and then you get rushed by them ;-) As we want to be able to upload GTK+ 2.10 to unstable as soon as possible, we'd like to move the current experimental 2.8 packages to unstable. Could you tell us what is exactly needed for that without breaking d-i? Do we need versioned conflicts with some packages? The gtk2.0+directfb packages _do_ have the same names as existing packages, so for that a coordinated upload is necessary. AFAIK versioned conflicts are of no use for d-i. We just have to make sure that udebs that depend on the cairo and gtk libs are rebuilt ASAP after the new libs hit unstable. Here come the fruits of our work, the udebs will be partly built at the same time as the regular debs enter unstable ;-). the only thing which needs building, AIUI is cdebconf(-gtk). The rest are already ok. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: GTK+ 2.8.18 experimental debs/udebs
On 03/07/06, Eddy Petrişor [EMAIL PROTECTED] wrote: I also remember that on sven's macintosh (which should be similar o yours), back at extremadura, we had to boot textua, run gtk-demo, close gtk-demo, and then run the graphical installer as we were experiencing a system crash very similar to the one you described. Could you try this latest trick too ? I hope to try this tonight in spite of real life engagements I have scheduled for tonight. Where from can I get the gtk-demo app source? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: GTK+ 2.8.18 experimental debs/udebs
The new packages: http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/libs/ It seems nobody complained about this, but the permissions are wrong for this. I will fix them tonight when I get home. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GTK+ 2.8.18 experimental debs/udebs
On 30/06/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: I also added a strace for a regular (no video=ofonly) session at the same place. I looked at both dfb and strace logs and this could be bug #342053 : have you tried booting with the textual frontend, adding no-hardware to directfbrc and then starting the graphical frontend ? It is not that bug, I have added no-hardware and tested and the crashing still happens. I don't remember if I said this before, ofonly also crashes[1] I also remember that on sven's macintosh (which should be similar o yours), back at extremadura, we had to boot textua, run gtk-demo, close gtk-demo, and then run the graphical installer as we were experiencing a system crash very similar to the one you described. Could you try this latest trick too ? I hope to try this tonight in spite of real life engagements I have scheduled for tonight. [1] http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/iso/ppc-gtk-2.8-crash-ofonly.log -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A web-based frontend for the cdebconf
On 03/07/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: does anyone has a better candidate than webfs ()? I think thttpd[1] has also cgi support, you might want to look at that, too. It is a little bit bigger, but probably, as an udeb, it can be smaller. [1] http://packages.debian.org/unstable/web/thttpd -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: A web-based frontend for the cdebconf
On 03/07/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: does anyone has a better candidate than webfs ()? I think thttpd[1] has also cgi support, you might want to look at that, too. It is a little bit bigger, but probably, as an udeb, it can be smaller. Good advice : i installed thttpd (which is said to support CGI) and the executable file's size is ~70 KB only : ATM i think this is really the best option. Glad to be helpful. As a self contained web server is doing a great job on my router, as seen until now ;-) . -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GTK+ 2.8.18 experimental debs/udebs
On 30/06/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: Eddy Petrişor wrote: I also added a strace for a regular (no video=ofonly) session at the same place. Note that I ran a session with video=ofonly which also crashed, but I didn't made a trace. I looked at both dfb and strace logs and this could be bug #342053 : have you tried booting with the textual frontend, adding no-hardware to directfbrc and then starting the graphical frontend ? No I didn't. But this seems rather odd because I had the graphical installer running on my laptop with old libs. Is there any reason to expect regressions? I also remember that on sven's macintosh (which should be similar o yours), I think he had an older model, mine is PowerBook5,2, iirc, his was PowerBook4,6 or something like that. Sven? back at extremadura, we had to boot textua, run gtk-demo, close gtk-demo, and then run the graphical installer as we were experiencing a system crash very similar to the one you described. Could you try this latest trick too ? I will tonight, but I fear there is something else here... Sven, could you try the image I made[1] and see if is working for you? [1] http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/iso/miniiso2.8.18-powerpc/20060629mini.iso -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: GTK+ 2.8.18 experimental debs/udebs
On 30/06/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: Eddy Petrişor wrote: I've build the image with gtk 2.8.18...and it crashes. A friend tested the PPC mini ISO on a minimac and he had to boot with video=ofonly to prevent the installer from crashing, and colours were messed up as usual. Wasn't that only for the mouse cursor? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: GTK+ 2.8.18 experimental debs/udebs
On 29/06/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: Eddy Petrisor wrote: Done, they are available on my router[1], in case anybody else wants them. Man, the compilation needed about 1.5GB free disk space.. [1] http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/libs/gtk/2.8.18-4/ Good work! :) Any chance to build a ppc miniiso with updated dfb, cairodfb and gtkdfb libs ? there are some still open bugs (#342053, #338191, #341770) for cdebconf-gtk related to PPC architecture that i sustect are related to older libraries. I will try tonight, but I have other real life things to do tomorrow, so I can say for sure. On Friday I will leave Timişoara, so I am not sure if I can sqeeze it before the weekend. What's sure is that'll try to do it ;-) Also, it would be interesting to know wheter bug #373253 shows up on AMD64 only or also with other architectures different from i386. AFAIR, none of the working images for powerpc had this issue. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: GTK+ 2.8.18 experimental debs/udebs
As a summary, now we have on powerpc: I have recompiled the cairo packages yesterday for powerpc tried to the cairo packages are at: http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/libs/cairo/libcairo-1.1.10-2/ and gtk+2.0 packages: [1] http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/libs/gtk/2.8.18-4/ -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GTK+ 2.8.18 experimental debs/udebs
On 29/06/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: ok, there is no hurry: i was just curious to know wheter some ppc bugs were gone with newer libs. I am also curious about this. Probably some publicity on -powerpc, after I have build some images might be a good way to attract testers. i realy hope #342053, which is the most severe, has disappeared in DFB 0.9.24. I would put my money on it, but we'll see. For sure new libraries will need some testing. If others don't object, then, after te build of the images, I shall make a call for testers for the iso image with new libriaries.. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GTK+ 2.8.18 experimental debs/udebs
On 29/06/06, Frans Pop [EMAIL PROTECTED] wrote: On Thursday 29 June 2006 10:19, Eddy Petrişor wrote: If others don't object, then, after te build of the images, I shall make a call for testers for the iso image with new libriaries.. Please test it extensively yourself first and let us know the results. Wider testing is only useful if there are no obvious issues. Right, will do. To be honest, I was thinking more about things like hardware diversity when I said that (for instance, I have never seen #342053 on my laptop ;-). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
How new are the news?!
Hello all, I wanted to point out that the translation status page could use some cleanup in the news section. The last entry is from 2004/04/27 which, in my universe does not count as news ;-). Probably is wise if the news section is shortened and (if desired) the old entries are moved in another place. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: GTK+ 2.8.18 experimental debs/udebs
Sadness is filling my body as I have seen yet another G-I on powerpc :(( On 29/06/06, Eddy Petrişor [EMAIL PROTECTED] wrote: On 29/06/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: ok, there is no hurry: i was just curious to know wheter some ppc bugs were gone with newer libs. I am also curious about this. Probably some publicity on -powerpc, after I have build some images might be a good way to attract testers. i realy hope #342053, which is the most severe, has disappeared in DFB 0.9.24. I would put my money on it, but we'll see. I've build the image with gtk 2.8.18...and it crashes. For sure new libraries will need some testing. If others don't object, then, after te build of the images, I shall make a call for testers for the iso image with new libriaries.. This will have to wait. Image and sdterr output: http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/iso/ The new packages: http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/libs/ -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: GTK+ 2.8.18 experimental debs/udebs
On 30/06/06, Eddy Petrişor [EMAIL PROTECTED] wrote: Sadness is filling my body as I have seen yet another G-I on powerpc :(( [snip] I've build the image with gtk 2.8.18...and it crashes. For sure new libraries will need some testing. If others don't object, then, after te build of the images, I shall make a call for testers for the iso image with new libriaries.. This will have to wait. Image and sdterr output: http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/iso/ I also added a strace for a regular (no video=ofonly) session at the same place. The new packages: http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/libs/ -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: GTK+ 2.8.18 experimental debs/udebs
On 27/06/06, Josselin Mouette [EMAIL PROTECTED] wrote: Le mardi 27 juin 2006 à 00:29 +0200, Davide Viti a écrit : I started testing libgtk+2.0-directfb0-udeb_2.8.18-3_i386.udeb I've rebuilt gtk+ against a fixed glib, the correct cairo, and without all loaders except the PNG one. Same place: http://people.debian.org/~joss/packages/ I have recompiled the cairo packages yesterday for powerpc tried to compile the gtk packages, but it seems that the compialtion of this package needs about 1GB+ on the machine, I had issues with this. I will try again tonight. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: [g-i] new screenshots available
On 6/24/06, Davide Viti [EMAIL PROTECTED] wrote: On Sat, Jun 24, 2006 at 07:01:14AM -0700, Damodharan Rajalingam wrote: Hmm.. TSCu_Paranar is the font which I use in my Debian box and dont have any problem with it. Also there was a screenshot sent when Tamil was announced a supported language. The tamil font rendering in that screenshot was fine. Is it because of font stripping? Can we try with unstipped font? TSCu_Paranar is not stripped; problem is that freefont contains some glyphs in the U0b80-U0bff range and somehow has priority over Tamil fonts. Here how Tamil looks like using TSCu_Paranar. http://www.webalice.it/zinosat/ta.png I'm about to file a bug agains greefont. Thanx for reporting the problem. AFAICT this is the exact problem that I have observed in January because freefont had higher priority than all other in the G-I environment. I honestly don't think is a font issue, but a G-I issue. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: where to input bugs for gtk d-i?
On 6/23/06, Luis Matos [EMAIL PROTECTED] wrote: where can i submit some wish list bugs? cdebconf-gtk-udeb package. i re-installed yesterday my laptop, and i found some flaws in it, mostly non searchable box for languages, keyboards, etc. Where did you got from the image you tested. Some of the issues you are talking about may result from an old set of libraries. other thing is the theme choosed ... it is too square, maybe adopting the gnome default theme, that would be nice ... anyway, don't forget to answear to the question, please. There is already an intention to use a theme which is not gnome's default one, but that (the one I am talking about) is not at all square[1]. ;-) [1] http://lists.debian.org/debian-boot/2006/03/msg00910.html -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [g-i] new screenshots available
On 6/23/06, Viti Davide [EMAIL PROTECTED] wrote: Version: 20060501cvs-2 we use the udeb where many glyphs were removed for reducing memory footprint By anychance did you remove the glyphs of conjunts? :-) Do they fall on any of the following ranges? ranges=$ranges u0400:u052F # Cyrilic ranges=$ranges u0530:u058F # Armenian ranges=$ranges u0600:u06FF # Arabic ranges=$ranges u0700:u077F # Syriac ranges=$ranges u0D00:u0FFF # Malayalam, Sinhala, Thai, Lao, Tibetan ranges=$ranges u10D0:u10FF # Georgian ranges=$ranges u1EA0:u1EFF # Vietnamese ranges=$ranges u2100:u21FF # Letterlike, Number forms, Arrows ranges=$ranges u2200:u24FF # Mathematical, Techical, Control, OCR ranges=$ranges u2580:u259F # Block elements ranges=$ranges u3000:u30FF # CJK Symbols and Punctuation ranges=$ranges 65536:66002 # bn_* stuff ??? This reminds me of the include what we need, not exclude what we think we don't discution. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: gtk+-directfb packages: call for help
On 6/21/06, Davide Viti [EMAIL PROTECTED] wrote: We think there's everything you need to produce gtk+-directfb packages; it's not officially available yet, but packaging could be done anyway and would help the d-i team to start working on migrating from old libraries to new ones. Your help is thoroughly appreciated, TIA, As we talked on IRC, I have done some work on this issue. You can find what I have done[1]* in chronological order, so basically the most recent stuff is better. I will put the result of my work at the end of a day in separate directory (I know this is not version management ;-) ). [1] http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8transition/ * this machine sometimes blocks (I think it needs more cooling), so it might not be always accesible during the day, but when I arrive at home (around 7 o'clock in the evening, EET) I see this problem and restart it -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [directfb-dev] [g-i]GTK 2.8.18 with directfb support packages [was:Re: [g-i] Graphical installer and PPC systems]
On 6/20/06, Viti Davide [EMAIL PROTECTED] wrote: I have been fighting the whole Sunday to make a patch to the debian package gtk+2.0 so that it produces also a library with the directfb backend. why? The Gnome team will support the D-I team in producing the gtk+-directfb packages. I wanted to do so... and if I do it right, their work will be less. Until now I have maneged to make a separate set of targets in the debian/rules. I am still unsure about the way I should tell the build system to use the cairo.so file provided by the libcairo2-directfb-dev package (Dave, I fear that symlinks will be necessary for this file, too). Also I am not sure if the so file should not be called something like cairo-directfb.so, but am so unsure of these things that i think nothing should be done until this issue is calrified. Cairo is fine and, apart from the couple of things pointed out by Frans, package can be used for compiling gtk+: why should we delay things? why should the library be renamed? there's no such an issue and things have been handled properly via .pc file. have you checked the packages? (yes, I have) Would you please double check before sendig such messages to package maintainers and Multiple MLs? Now I think I got the general idea behind the build process of gtk+2.0 Debian package and I hope I will manage to get tonight the direcfb library to build. The library builds fine already (see [2] and [3]). The impression that the above messages left me is that the library used for the image was simply built from sources and added in the image via a tarball, not via a package. Am I not correct about this? Are you familiar with rebuilding all the libraries, the needed udebs and the mini.iso (I know you've rebuilt from source Dave's package)? have you managed to rebuild the new libraries and to create a g-i image based on those? if so, can you please provide a link to a ppc mini.iso based on the new libs? From one of your messages on d-boot ([1]) you say that current ppc g-i is broken, so shouldn't the crash be fixed before trying to use the new libraries or do you blindly assume that everythings will be magically ok with the new libs? Note: I feel much anger in your writing... I don't consider it my responasbility to fix those images, as I have said some time ago, I just have time for testing. In spite of that I wanted to see gtk+2.0 packages prepared to build directfb packages, too and hoped (not blindly) that the colour issues will disappear. your help is of course _very_ welcome, but to me it looks like you're doing the wrong thing at the wrong time: rebuilding the libs from scratch was very useful a while back (before and soon after Extremadura), when Mike needed the support for putting the dfb backend into cvs and at the time Attilio and myself spent alot of time and effort on doing this (see wikis). now what is needed is testing the packages produced by the official maintainers and not duplicating their work or delaying other people' work. Have they done this? Judging from what I know and seen in unstable, directfb gtk packages are still unsupported. Do such experimental packages exist? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: [directfb-dev] [g-i]GTK 2.8.18 with directfb support packages [was:Re: [g-i] Graphical installer and PPC systems]
On 6/20/06, Viti Davide [EMAIL PROTECTED] wrote: why? The Gnome team will support the D-I team in producing the gtk+-directfb packages. I wanted to do so... and if I do it right, their work will be less. I'm not sure they need help on packaging; I might be wrong but I don't think the gtk is not ready because of technical problems. my feeling on the #gnome-debian channel ws that they were busy with other stuff ATM and nobody had time for this. The library builds fine already (see [2] and [3]). The impression that the above messages left me is that the library used for the image was simply built from sources and added in the image via a tarball, not via a package. Am I not correct about this? You are right, but why do you need a debian package to test the patch? And again: have you tried to build a mini.iso based on the new set of Libraries? Wouldn't it make sense to get confident with it before concentrating On making an udeb you would not know how to use and test? Err, i thought your i386 tests made that. If I am doing the patching against the source package, then you could test it on i386, too. Is not like they are hacked udebs like Attilio was doing, so you can't use it. Note: I feel much anger in your writing... Yes, your message did piss me off very much indeed. You would delay things by saying i think nothing should be done until this issue is clarified because you have not clear how the libraries work. Note that it's perfectly ok having doubts, but before telling that in name of The d-I team to the package maintainer, you could discuss it on irc/ml and make Sure there are some problems. IMO it just confuses things. I got no clear response on that matter from gtk/gnome people on the debian channel, and also I have faced some issues that made me believe that having two cairo.so files in two different places could be problematic; indeed I might have been wrong by not carifying that I was talking about _my_ effort. I don't consider it my responasbility to fix those images, as I have said some time ago, I just have time for testing. Huh? What do you expect then? I said clearly I don't have time for hacking[1], just testing. I am not the ppc porter, nor a helper porter and I don't have the knowledge to claim any of those positions. Also, somebody else is the powerpc porter, AFAIK. You write on d-boot hey, g-I is fucked on ppc as you did with Cyrillic fonts, And then you expect within a couple of hours to find a hundred messages on your Mailbox asking for details or providing a fix? Err, the problem was observed on the 18th of May, and I was expecting, since then somebody to try to investigate and fix this since I clearly said I couldn't do it. I have said something about this on the list on the 30th of May. I know the wiki.d.o/D-I/GUI is not BTS, but I have been told that it would not be proper to add this issue to D-I/Today and since the builds were not integrated, IIRC in trunk for PPC, I thought BTS was not the proper place to report them. The dailies were not official, thus BTS seemd wrong to me. You lately reported that the crash on ppc happens because of the touchpad fix, but you Did not even try to revert the patch and see if that is the real problem... Are you sure[2]? snipped from wiki.d.o: [2006-06-14] ppc: the debconf interface crashes (the fix for [WWW] #372773 is working here) If that's your idea of testing things I'd say is not very effective and as a consequence it Takes a long time before problems get fixed. What part of [1] didn't you understand? In spite of that I wanted to see gtk+2.0 packages prepared to build directfb packages, too and hoped (not blindly) that the colour issues will disappear. The colour problem disappears: it's one of the first things we did In Extremadura. Sven built the libraries and a test image and, apart For a problem with some buttons (not used inside g-i), everything looks Beautiful. I don't think you need a deb to see thet. But the packages are needed nonetheless. now what is needed is testing the packages produced by the official maintainers and not duplicating their work or delaying other people' work. Have they done this? Judging from what I know and seen in unstable, directfb gtk packages are still unsupported. Do such experimental packages exist? No, it does not exist yet; cairo is a requirement. Unofficial cairo Could be used, true, but we'll need it officially anyway. That's what I am working on. I think the gnome team have all the skills needed to package gtk+-directfb, No doubt, but the time is pressing us; if we want a G-I for etch every day counts, even the energy spent now could have been spent a lot better. So IMO it's not a technical problem and of course you can try to do it Yourself, but I don't know how useful that would be. Dave announced the availablity of the cairo packages for quite a while and I haven't seen any move from the gtk/gnome team. The most
Re: [directfb-dev] [g-i]GTK 2.8.18 with directfb support packages [was:Re: [g-i] Graphical installer and PPC systems]
On 6/20/06, Frans Pop [EMAIL PROTECTED] wrote: On Tuesday 20 June 2006 16:50, Eddy Petrişor wrote: Dave announced the availablity of the cairo packages for quite a while and I haven't seen any move from the gtk/gnome team. The most important thing I learen in OSS is that you (yourself) are the best o scratch your own itch. That's what I think I am doing here. There is still some discussion about those packages and I reported at least one important needed change a few days ago. AFAIK we have not yet asked the Gnome people to do anything and that seems the smart thing to do until we _know_ the cairo packages are actually good. Probably the best to say that is actually someone who tries to use them or someone with enough library packaging experience. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: [directfb-dev] [g-i]GTK 2.8.18 with directfb support packages [was:Re: [g-i] Graphical installer and PPC systems]
On 6/20/06, Davide Viti [EMAIL PROTECTED] wrote: On Tue, Jun 20, 2006 at 05:50:31PM +0300, Eddy Petrişor wrote: following the wiki or using the build script committed in extremadura you should be able to build any of the g-i libs and that is what Sven, Attilio and myself did. Packaging the libraries, requires alot more; I just got the impression you are not confident with none of those (and I really do not mean to offend you anyhow by saying that) On the contrary, I have trust in your, Attilio's and Sven's work. I just want to take this to the next level. I don't think it helps a lot if I do something which has been done before, does it ;-) ? I got no clear response on that matter from gtk/gnome people on the debian channel, and also I have faced some issues that made me believe that having two cairo.so files in two different places could be problematic; indeed I might have been wrong by not carifying that I was talking about _my_ effort. I think it was not appropriate to involve two MLs and Dave to sort out your doubt: cairo has been successfully used to produce working images and that is enough to say it is working. Err, your are missing the point, you and everybody else used only one library the cairo.so which is build against -directfb. I am not sure that in the context of packaging gtk this is fesable because you have to specify the library used for linking ( -lcairo ), and that is provided for the single source package gtk+2.0 by two -dev packages which both have such a library. What I am NOT sure (and that was one of the points of my initial mail) is if this can be done for a package depending on both. You write on d-boot hey, g-I is fucked on ppc as you did with Cyrillic fonts, And then you expect within a couple of hours to find a hundred messages on your Mailbox asking for details or providing a fix? Err, the problem was observed on the 18th of May, and I was expecting, [cut, me saying wrongly that I have reported this the first time on the list on] the 30th of May. you reported the pb in [1] and the very same day I replied asking for infos Yes, you are right I didn't report it first time on the 30th, but still since the 22nd until the 25th still there are not a couple of hours. Read further. you did not provide until the IRC discussion when you unfairly accused Colin. Get your act together, read [1] which was sent on the 23rd. I CC-ed Colin on that in the hope I'll rasie his attention. you asked if I could reproduce [2] and obviousely that was reproducible since the fix came on the 26th. I have done that [3] on the 25th and also provided some screenshots. I asked you again to provide the infos and within 12 hours Colin merged ppc stuff from Frans' branch into trunk. On the 25th @ 21:06+3 when I said that Collin didn't cared about the issue. He himself has acknowledged that he didn't looked at the problem. He moved the build to the trunk on the 26th @ 14:05+1 And he didn't provided an explicit link to the images nor did he updated the links on the wiki. I am not acusing him, is just the information was missing. You said the problem was still present [2] (and I had to ask you the very same At the second round of images when I actually tested a wrong (old) image because of a wrongly setup cron job which couldn't overwrite a file made by the root user. Indeed that was my mistake and I have said that publicly. If I remember correctly I had dowloaded the wrong image by hand because of this lack of info. questions again [3]), and at the end it was you using the wrong image. That did piss me off even more and I think such failure reports are all but useful. Flashnews, nobody's perfect. I acknowldged that was my mistake. I hope you don't have to near somebody digging up dirt from your past to pour it in your head. Especially when you have acknowledged publicly you were wrong. Still I fail to see what does this have to do with my work on the gtk packages. Are you sure[2]? right, my fault (i'm sure some posts you bamed it on that patch though) Can you backup that afirmation? taking into account the omissions before I suspect your memory is cheating on you. I don't remember saying anything about any patch. The touchpad issue was reported by someone else while I had said that the G-I didn't started on ppc suspecting some input problems because of the messages I could see between the crashes. If that's your idea of testing things I'd say is not very effective and as a consequence it Takes a long time before problems get fixed. What part of [1] didn't you understand? it's crystal clear. reporting problems the way you did, especially when many ppl are on different arches, does not help at all IMO, but A lot of hand waving... I said PPC, what does diffrent arches has anything to do this? go on that way if you think I'm wrong. I have said I will try to test the images weekly. I expected somebody else (maybe the officail porter) to do
[g-i]GTK 2.8.18 with directfb support packages [was: Re: [directfb-dev] [g-i] Graphical installer and PPC systems]
On 5/25/06, [EMAIL PROTECTED] [EMAIL PROTECTED] wrote: Mike Emmel wrote: I think a lot of the issues were actually 64 bit stuff thats recently been fixed any chance you could try 25.1 ? or a cvs build just to see. I believe the problems were down in the directfb layer so it would be nice to know if its there or higher up in gdk/cairo. What we really need is a commited developer with a PowerPC to trace this down thats the main problem. I've updated the GTKDFB howto [1] to build latest GTKDFB libraries: could someone with a PPC box try to download and compile DFB 0.9.25.1 and then follow the steps explained in the wiki tutorial to build GTKDFB PPC libraries ? I have been fighting the whole Sunday to make a patch to the debian package gtk+2.0 so that it produces also a library with the directfb backend. Until now I have maneged to make a separate set of targets in the debian/rules. Still there are some problems because in the Debian package it seems that the configure* files have also some .orig counterparts and I haven't figured yet what is the flow of the things during the build for these files. (directfb gdk-target is still not visible, because the configure* files are rewritten - this is the most recent info I have, but I had to blindly try to compile the package and went on a wrong track for some time causing me to waste about 4-5 hours). I am still unsure about the way I should tell the build system to use the cairo.so file provided by the libcairo2-directfb-dev package (Dave, I fear that symlinks will be necessary for this file, too). Also I am not sure if the so file should not be called something like cairo-directfb.so, but am so unsure of these things that i think nothing should be done until this issue is calrified. Now I think I got the general idea behind the build process of gtk+2.0 Debian package and I hope I will manage to get tonight the direcfb library to build. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: This is getting ridiculous ...
On 6/17/06, Wouter Verhelst [EMAIL PROTECTED] wrote: And what has that to do with anything ? I have no interest in going into this hate-game, i want the situation solved, and be able to commit as any normal DD would. You have that already. There are 1000 DD's, and only a small portion of those has commit access to the d-i repository. any normal DD, therefore, does not have commit access to the d-i repository. And how many of them cannot commit despite wanting to do it ? Not a clue. == I don't know, aka lack of knowledge, aka diletantism (I am not sure if this word is too strong) No interest either == I don't care, aka ignorance -- how is it any relevant? == lack of touch with reality Please, don't be ridiculous. Sven is a person who: - has the techinical skills to support the powerpc arch - never abused his commit rights (AFAIK, and nobody has ever argumented against this) - has a communication problem on which he is definetly working on and the progress is visible - has been involved in Debian for a long time - wants to contribute to D-I Asking how is it relevant if people in Debian (and I mean DDs) which wish for D-I access and don't have it may not exist makes me believe that you are not talking seriousely. I personally don't think such people existst, or if they do, they are involved since only quite recent times or they do not feel they need it. The commit right was administered very well until this event, IMHO. Heck, there are translators which have access to D-I SVN (me being one of them), and Sven's has been revoked? Yes, he might have said that he will not do D-I work, but he has changed his mind. Is that a reason not to give it back? (and please don't argument with ridicule things such as a ping pong of I want it/I don't want it on Sven's side) Is there any TECHNICAL reason for him not being given back SVN access? Are there ANY reasons to believe that he will do submersive acts by using his SVN access? Isn't he allowed to be given a second try? I feel that Frans has taken a cunning course of actions in what seems to be a personal vendeta, and he is taking advantage of the fact that people have been annoyed at different times by Sven, and by his rights over the D-I repository. Frans, I really don't care about your oppinon now, you have became a small person in my eyes (thing of which you might not care, but I have granted you the right to a second mistake). I have told you in a private mail in the past that you are making a mistake in the way you are handling the situation, that you are overreacting, while you have hidden behind the techincal comitee's decision and then behind AJ's. None of these two has never acknowledeged, AFAIK, the social problem behind the whole Sven's SVN access rights issue, and I feel this is a failure. Why? Because ALL of the current situation is based on personal feelings. I said cunning, and I will say it again, because you _are_ cunningly provoking Sven while he is such an easy target because of his justified, IMO, feeling that he has been incorrectly been ruled against and he is irritated by this. Yes, Sven is not a man without a flaw, but he knows it and he is working on it, OTOH, you, Frans are also not either without flaw, but you are a lesser man because you are not making your best to try to put animosity behind yourself and let your personal feelings cloud your judgement and affect D-I work on the powerpc port. I have been silent about this issue for quite some time in the hope that things will not degenerate, but I see that pride is higher positioned than wisdom in some places within Debian. Sven, I accuse you of being at guilt about the IRC conversation that triggered this whole thread, but I feel that you are indeed sincere when you say that you feel this is ridiculous. You have been such an easy catch for Frans that I feel sorry for you for not being able to step away from this whole issue. I know is hard, but you are doing yourself a disfavour and, although others might see your break from this issue as a win on their side, you could be able to improve your own state of mind. You have a great deal of leaning at keeping yourself out of trouble (and that might be a from some POV-s a good thing). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2 - time for gtk+-directfb
On 6/14/06, Dave Beckett [EMAIL PROTECTED] wrote: -BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Cairo 1.1.8 snapshot was just released so I've made a new set of experimental debs for it at: http://download.dajobe.org/debian/experimental/ If any powerpc people care, I build both 1.1.6 and 1.1.8 libcairo packages for powerpc[1]. Hopefully during this weekend I will have also gtk packages built for D-I (but I fear that the packaging scheme for gtk is quite compilcated for my skills or understanding - I didn't figured out how things move in the package, so I could apply the patch needed to get directfb support in the 2.8.* version of gtk). I will try to patch gtk and make directfb packages during this weekend, but if not successful, I will probably fall back to tarballs. [1] http://eddyp.homelinux.net:8080/eddy/g-i/gtk2.8-ppc/libs/cairo/ -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Bug#372773: [powerpc][newwolrd][g-i] cdebconf-gtk interface crash (daily image)
On 6/15/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: The fix given in #372773 is working for me, too. This seems weird, the G-I daily worked fine for me around the 18th of May. Is this due to directfb 0.9.24? uhm.. afaik daily images are still built with DFB 0.9.21: migration to DFB 0.9.24, GTKDFB 2.8.17 and cairodfb 1.1.6/8 has to happen all at once. Then why this sudden change in behaviour? This issue related to touchpads was not present around the 18th of May. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2
On 6/13/06, Eddy Petrişor [EMAIL PROTECTED] wrote: Or are you refering to the fact that the package is new in experimental? It's not new yet. I'm waiting for somebody here to tell me to upload it - and whether to set it for experimental or unstable. I will build it for powerpc and see if I can build a working image with this udeb and a (hacked?) gtk one. Can someone do this? It seems that my ISP @ home has decided that it was too much tie for me having their service running without problems and now I don't have my i-net connection working. I don't know when it will be fixed, but they have a bad reputation of answering this kind of things quite slowly. Sven? Could you try? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
[powerpc][newwolrd][g-i] cdebconf-gtk interface crash (daily image)
Hello all, I have reported this issue for about a week and I don't know yet if is related to #372773, but the G-I interface (powerpc daily image) crashes and I haven't seen anybody interested in fixing this issue. Can somebody confirm/infirm this? Tests have been ran on a PowerBook5,2. I have added this info on the /GUI wiki page. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: currently broken udebs in etch
On 6/13/06, Eddy Petrişor [EMAIL PROTECTED] wrote: Probably non D-I udebs should be on this list, too. (I personally have fixed this for ppp-udeb and hope to publish it tonight, after some testing). I am sorry, I couldn't make this happen due to some problems with the internet connection I use at home. I don't know when this will be fixed as it depends totally on my ISP. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Bug#372773: [powerpc][newwolrd][g-i] cdebconf-gtk interface crash (daily image)
On 6/14/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: Eddy Petrişor wrote: On 6/14/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: Eddy Petrişor wrote: On 6/14/06, Attilio Fiandrotti [EMAIL PROTECTED] wrote: Eddy Petrişor wrote: Hello all, I have reported this issue for about a week and I don't know yet if is related to #372773, but the G-I interface (powerpc daily image) crashes and I haven't seen anybody interested in fixing this issue. Can somebody confirm/infirm this? Tests have been ran on a PowerBook5,2. I have added this info on the /GUI wiki page. as you don't provide many details , i can only suggest to boot in NEWT, add disable-module=linux_input to /etc/directfbrc and then try going graphical. If the crash still persists, try adding also no-hardware to directfbrc (it may be bug #342053) I have started the G-I previousely on this hardware (remember the missing cyrillic fonts issue), but I think the issue I am hitting is #372773. I knew about the input issue yesteday, but I didn't knew the exact needed changes that were needed. As my i-net connection was not working, I wasn't able to test with this modification as I didn't knew the needed modification. Today I wrote this and will try again this evening. ok, let me know if it works Is it enough for me to add the line disable-module=linux_input in /etc/directfb in the D-I environment and restart the debconf interface (via GTK) or do i need a rebuilt rootskel-gtk udeb? adding line(s) is enough: directfbrc is parsed everytime a DFB-based application, like the graphical frontend, is started The fix given in #372773 is working for me, too. This seems weird, the G-I daily worked fine for me around the 18th of May. Is this due to directfb 0.9.24? I tried with the known trick: - boot with install DEBIAN_FRONTEND=newt - switch to VT2, export DEBIAN_FRONTEND=gtk - run debian-installer (now after adding the line disable-module=linux_input in /etc/directfbrc) -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Re: Request for a DFB version of libcairo2
On 6/12/06, Dave Beckett [EMAIL PROTECTED] wrote: I've made another cairo 1.1.6-1 experimental deb set, this time there's a new libcairo2-directfb-dev which can be parallel installed with the libcairo2-dev and answers to 'pkgconfig cairo-directfb --libs' etc. It's in the same place as the other packages: http://download.dajobe.org/debian/experimental/ Don't forget as there are new deb packages here, there's at least a month to wait for the NEW queue to process them before you can use them after I upload them to debian. Is that applicable to binary packages? I thought it was only for source packages... Or are you refering to the fact that the package is new in experimental? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: currently broken udebs in etch
On 6/13/06, Patrick Winnertz [EMAIL PROTECTED] wrote: ATM Debian-Edu is not installable because the change from prebaseconfig.d to finish-install.d since we use etch as basis of our CDD. This here is a list of the packages which still install scripts into the /usr/lib/prebaseconfig.d/ dir in etch and because of that they don't work. localechooser base-installer brltty-udeb clock-setup di-utils save-logs mdcfg-utils tzsetup-udeb user-setup-udeb Probably non D-I udebs should be on this list, too. (I personally have fixed this for ppp-udeb and hope to publish it tonight, after some testing). -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2
On 6/13/06, Dave Beckett [EMAIL PROTECTED] wrote: Eddy Petrişor wrote: On 6/12/06, Dave Beckett [EMAIL PROTECTED] wrote: Don't forget as there are new deb packages here, there's at least a month to wait for the NEW queue to process them before you can use them after I upload them to debian. Is that applicable to binary packages? I thought it was only for source packages... No, it's both. See flamewars on various debian lists re just new binary debs. Or are you refering to the fact that the package is new in experimental? It's not new yet. I'm waiting for somebody here to tell me to upload it - and whether to set it for experimental or unstable. I will build it for powerpc and see if I can build a working image with this udeb and a (hacked?) gtk one. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
Bug#279855: Blocked by missing PPP support
On 6/11/06, Filipus Klutiero [EMAIL PROTECTED] wrote: Note that this is blocked for Etch by the fact that [decent] PPP support is no more present. I have been working recently on the PPPoE issue; is almost finished but I still face two problems. For more information see threads related to: http://lists.debian.org/debian-boot/2006/05/msg00440.html http://lists.debian.org/debian-boot/2006/05/msg01267.html -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Doubt with GTK FB.....
On 6/12/06, Sandesh Sethia [EMAIL PROTECTED] wrote: Hello, This is my first post on this group. I am a novice programmer who wanted to know that can I use GTKDFB for creating a PNG image of any window (browser especially) delivered to the GTKDFB toolkit through a C++ program ? I do not want to use X window system and so was looking for GTK direct frame buffer access, but certain positngs on the web suggest that GTKFB is to be used on embedded devices only and that it is not a good tool to use on desktop applications. is it true? Please do let me know at the earliest as i m in a crunch situation. You might be better asking questions about DirectFB on the directfb-dev mailing lists (see CC) -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2
On 6/9/06, Viti Davide [EMAIL PROTECTED] wrote: Indeed, there is no need for two -dev packages, is there? Why would you force ppl interested only in the x11 version to install the directfb one as well? Only two pc files and two libs, right? No: include files differ. Say what? The source is the same... that's what I said about three mails ago. AFAICT there shouldn't be any difference beween the headers because they are the same headers. Are there any? -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2
On 6/9/06, Viti Davide [EMAIL PROTECTED] wrote: Say what? The source is the same... But compiled with different flags that's what I said about three mails ago. Then what you said three mails ago is wrong. I was asking then if they were the same... and nobody said that I was wrong about it, so I assumed my statment was right... AFAICT there shouldn't be any difference beween the headers because they are the same headers. Are there any? Have you checked it before stating they're the same? I don't have access to a linux machine now to test that. Have you checked? Discaimer: this might feel a little combative, but is not. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2
On 6/9/06, Viti Davide [EMAIL PROTECTED] wrote: Do you think I would have said that without having checked before? You haven't made it easy to let us understand that you did. You don't need a linux machine to check that; grab the -dev.deb and -dev.udeb in the link provided by Dave. Ok, thanks for clarifying that out; after all, that's what I was asking initially; there is the cairo-features.h files which is diferent (generated). Just hope all this unneeded noise will not confuse Dave. So the conclusion is that, because the cairo-features.h and the cairo libraries have the same name, there should be two -dev packages after all... each with its own pc file, lib file(s) and headers. These should be installable on the same system... (or am I wrong again?), thus the cairo headers will probably be placed under distict directories under usr/include (same for libs, but adapted and probably the directfb library can be renamed). /me thinks: all this just because Debian is a binary based distribution... -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2
On 6/9/06, Viti Davide [EMAIL PROTECTED] wrote: On 6/9/06, Viti Davide [EMAIL PROTECTED] wrote: Do you think I would have said that without having checked before? You haven't made it easy to let us understand that you did. Here's a snippet of the message you're referring to: Only two pc files and two libs, right? No: include files differ. I said they differ; if you don't trust what I said is up to you to double check it yourself. For me history began a little bit earlier :-), more exactly, at this snippet: -- If that is true, then the headers should not be different and the information in the pc can point to the two cairo libs which will be in /usr/lib/ no problems. there will be two pc files: cairo.pc (already created) cairo-directfb.pc - this can be created inside debian anc copied in the right place via debian/libcairo2-directfb-dev.install with a line like debian/cairo-directfb.pc /usr/lib/pkgconfig the same way it's done in the gtk+2.0-directfb package -- (With my poor knowledge of packages and policy) I would use /usr/include/directfb which is already used by directfb itself. I think that is a polution of the directfb space, I would go for usr/include/cairo-directfb -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2
Hmm. There seems to be some misunderstanding here: [snip] I think that also means that we are still missing one other package as AFAIK a -dev package normally depends on a corresponding regular library package. That would mean a package: libcairo2-directfb_1.1.6-1_i386.deb Indeed. So, we'd have: - libcairo2_1.1.6-1_i386.deb - libcairo2-dev_1.1.6-1_i386.deb - libcairo2-directfb_1.1.6-1_i386.deb - libcairo2-directfb-dev_1.1.6-1_i386.deb - libcairo2-directfb-udeb_1.1.6-1_i386.udeb - libcairo2-doc_1.1.6-1_all.deb Don't forget the source packages, so I can compile and test them on powerpc ;-) The regular and -directfb lib and -dev packages would need to be co-installable on a regular Debian system. Yes, that is an important point, solvable by placing things under /usr/lib/libcairo-directfb as the policy indicates. (Anybody please correct me if any of this is incorrect!) Thanks for clearing that up, I was reading the first messages of the thread and was about to send a clarification mail. The list you sent seems complete. -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2
On 6/8/06, Dave Beckett [EMAIL PROTECTED] wrote: The regular and -directfb lib and -dev packages would need to be co-installable on a regular Debian system. That's not possible since both -dev packages have to provide the same file /usr/lib/pkgconfig/cairo.pc and conflict, so that 'pkgconfig cairo --libs' No, I don't think that's right, you will have somthing like is done for gtk packages[1][2]; I think the packages should provide cairo.pc and cairo-directfb.pc As is done there: usr/lib/pkgconfig/gdk-directfb-2.0.pc usr/lib/pkgconfig/gtk+-directfb-2.0.pc are provided by libgtk+2.0-directfb-dev while usr/lib/pkgconfig/gdk-2.0.pc usr/lib/pkgconfig/gdk-pixbuf-2.0.pc usr/lib/pkgconfig/gdk-pixbuf-xlib-2.0.pc usr/lib/pkgconfig/gdk-x11-2.0.pc usr/lib/pkgconfig/gtk+-2.0.pc are provided by libgtk2.0-dev. Am I missing something? [1] http://packages.debian.org/cgi-bin/search_contents.pl?searchmode=filelistword=libgtk%2B2.0-directfb-devversion=unstablearch=powerpcpage=1number=all [2] http://packages.debian.org/cgi-bin/search_contents.pl?searchmode=filelistword=libgtk2.0-devversion=unstablearch=powerpcpage=1number=all etc. work in dependent packages. There's no 'pkgconfig cairo-directfb --libs'. It could be made so that libcairo2 libcairo2-directfb are co-installable (under /usr/lib/libcairo-directfb) but the libcairo2-dev and libcairo2-directfb-dev must conflict/replace each other. Dave -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED] -- Regards, EddyP = Imagination is more important than knowledge A.Einstein -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Request for a DFB version of libcairo2
On 6/8/06, Dave Beckett [EMAIL PROTECTED] wrote: Eddy Petrişor wrote: On 6/8/06, Dave Beckett [EMAIL PROTECTED] wrote: The regular and -directfb lib and -dev packages would need to be co-installable on a regular Debian system. That's not possible since both -dev packages have to provide the same file /usr/lib/pkgconfig/cairo.pc and conflict, so that 'pkgconfig cairo --libs' [snip] Am I missing something? yes this change could be made, but this would require a patch to be made to gtk+2.0-directfb (udeb)'s configure to use pkgconfig cairo-directfb rather than pkgconfig cairo. Is that correct? Hmm, thining again at the problem... We are using now the same source for building the -directfb and the regular cairo, right? The compilation of the libs is made from exactly the same code, but is just configured differently, right? If that is true, then the headers should not be different and the information in the pc can point to the two cairo libs which will be in /usr/lib/ no problems. Thus, probably the packages will exclude libcairo-directfb-dev because the relevant info will be in libcairo-dev Please, someone correct me if I am wrong (because I wonder if strange things will happen like confusion when a package will try to link to either of the libs?!) -- Regards, EddyP = Imagination is more important than knowledge A.Einstein
