[opensuse-factory] Tumbleweed - Review of the Week 2015/25
Dear Tumbleweed users and hackers, During this week, Tumbleweed has been undergoing major work, which resulted in only one minor snapshot being released into the wild (20150612 hit the mirrors on Monday). Most changes were further preparations for GCC 5, which brings me to the actual, big news for this week: * GCC 5 has been checked in to be the default compiler collection for Tumbleweed from now on. Due to the nature of a compiler switch, the entire distribution (> 8000 source packages) are being rebuilt. A lengthy process that is currently still ongoing. As expected, the number of build failures went up a lot (as described in the past, the staging areas only rebuild a subset of packages, namely the ones that are shipped on the DVD; doing the full repository in stagings would be unrealistic, ,due to build power constraints). So, beloved hackers, your help is now needed to get the number of failed packages down again. In the last couple days I tried to mark the failures in the status screen [0] with some information on what could go wrong. Start your engines, branch as much as you can, fix the builds and submit them back to the devel projects and back into Tumbleweed. Let's show the world how fast we can get this back to 'normal' numbers of failures. Christian was kind enough to post in http://lists.opensuse.org/opensuse-factory/2015-06/msg00281.html the most common issues you might encounter and some word of advice to get the issues fixed. Also, a lot of upstreams have been fixing their codes already, so an update might be worthy to look at. Last, but not least, Fedora also performed the switch in preparation for their next release (F23) and you might as well borrow patches from there. Should you have problems with some packages, reach out to the larger community for help. [0] https://goo.gl/7EuFdQ Cheers, and always remember: have a lot of fun Dominique -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Sat, Jun 20, 2015 at 8:26 AM, Dominique Leuenberger a.k.a. Dimstar <dimstar@opensuse.org> wrote:
Christian was kind enough to post in http://lists.opensuse.org/opensuse-factory/2015-06/msg00281.html the most common issues you might encounter and some word of advice to get the issues fixed. Also, a lot of upstreams have been fixing their codes already, so an update might be worthy to look at. Last, but not least, Fedora also performed the switch in preparation for their next release (F23) and you might as well borrow patches from there.
I just looked at the suspiciuos failures of apps linking to libusb.. This is a bug introduced by https://build.opensuse.org/package/view_file/hardware/libusb-compat/libusb-c... and has nothing to do with the compiler switch. That patch is going to break something else sooner or later anyway all to please a buggy proprietary tool..:-| -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Sat, Jun 20, 2015 at 8:19 PM, Cristian Rodríguez <crrodriguez@opensuse.org> wrote:
I just looked at the suspiciuos failures of apps linking to libusb.. This is a bug introduced by https://build.opensuse.org/package/view_file/hardware/libusb-compat/libusb-c... and has nothing to do with the compiler switch. That patch is going to break something else sooner or later anyway all to please a buggy proprietary tool..:-|
Christian, indeed, Stanislav and me came to the same conclusion on Friday - that patch will temporarily be disabled until Stanislav can find a better approach of fixing this - hopefully next time around without the negative impact. For now, the biggest worry I have in Tumbleweed is the fact that p11-kit no longer passed the test suite (make check, which, sadly enough, is not enabled in the package). It seems to have issues with libffi, resulting in a lot of fails, as soon as certificates are involved (most of the java failures, KDE3 failures, many more) This has been logged as https://bugzilla.opensuse.org/show_bug.cgi?id=935510 Dominique -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
I just looked at the suspiciuos failures of apps linking to libusb.. This is a bug introduced by https://build.opensuse.org/package/view_file/hardware/libusb-compat/libusb-c... and has nothing to do with the compiler switch. That patch is going to break something else sooner or later anyway all to please a buggy proprietary tool..:-|
I am sure this package is not involved. I made a patch to what actually causes the failures in file named "libusb": https://build.opensuse.org/request/show/312876 -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Sun, Jun 21, 2015 at 4:50 AM, Ilya Chernykh <anixxsus@gmail.com> wrote:
I just looked at the suspiciuos failures of apps linking to libusb.. This is a bug introduced by https://build.opensuse.org/package/view_file/hardware/libusb-compat/libusb-c... and has nothing to do with the compiler switch. That patch is going to break something else sooner or later anyway all to please a buggy proprietary tool..:-|
I am sure this package is not involved. I made a patch to what actually causes the failures in file named "libusb": https://build.opensuse.org/request/show/312876
In this sample, you're right: libusb is not involved... but if you check the status overview [0] then this one was also not tagged to be affected by libusb at all. The patch in libusb-compat keeps on being the source for a problem; SR#312910 currently disables this patch, giving Stanislav the time to find an actual fix to the problem. Cheers, Dominiquee [0] https://goo.gl/7EuFdQ -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
Cristian Rodríguez wrote:
I just looked at the suspiciuos failures of apps linking to libusb.. This is a bug introduced by https://build.opensuse.org/package/view_file/hardware/libusb-compat/libusb-c... and has nothing to do with the compiler switch. That patch is going to break something else sooner or later anyway all to please a buggy proprietary tool..:-|
I spent several hours on my patch in Friday. Now I know what is wrong: The wrapper hides libusb-1.0 symbols from the application. But it does not do it for static inline functions in libusb.h. They are used in two libusb0 functions. If they are used, application fails to link. I have to rewrite the patch: Generate custom version of libusb.h header referring to wrapper functions. -- Best Regards / S pozdravem, Stanislav Brabec software developer --------------------------------------------------------------------- SUSE LINUX, s. r. o. e-mail: sbrabec@suse.com Lihovarská 1060/12 tel: +49 911 7405384547 190 00 Praha 9 fax: +420 284 084 001 Czech Republic http://www.suse.cz/ PGP: 830B 40D5 9E05 35D8 5E27 6FA3 717C 209F A04F CD76 -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Sat, Jun 20, 2015 at 1:26 PM, Dominique Leuenberger a.k.a. Dimstar <dimstar@opensuse.org> wrote:
Dear Tumbleweed users and hackers,
During this week, Tumbleweed has been undergoing major work, which resulted in only one minor snapshot being released into the wild (20150612 hit the mirrors on Monday). Most changes were further preparations for GCC 5, which brings me to the actual, big news for this week:
* GCC 5 has been checked in to be the default compiler collection for Tumbleweed from now on.
Due to the nature of a compiler switch, the entire distribution (> 8000 source packages) are being rebuilt. A lengthy process that is currently still ongoing. As expected, the number of build failures went up a lot (as described in the past, the staging areas only rebuild a subset of packages, namely the ones that are shipped on the DVD; doing the full repository in stagings would be unrealistic, ,due to build power constraints).
So, beloved hackers, your help is now needed to get the number of failed packages down again. In the last couple days I tried to mark the failures in the status screen [0] with some information on what could go wrong. Start your engines, branch as much as you can, fix the builds and submit them back to the devel projects and back into Tumbleweed. Let's show the world how fast we can get this back to 'normal' numbers of failures.
Christian was kind enough to post in http://lists.opensuse.org/opensuse-factory/2015-06/msg00281.html the most common issues you might encounter and some word of advice to get the issues fixed. Also, a lot of upstreams have been fixing their codes already, so an update might be worthy to look at. Last, but not least, Fedora also performed the switch in preparation for their next release (F23) and you might as well borrow patches from there.
Should you have problems with some packages, reach out to the larger community for help.
So far I identified three packages that are responsible for a big numnber of fallouts: * libusb-compat: a patch that attempted to fix closed source softare has some negative effects. An SR, disabling the patch for now, is pending against the devel prj (SR#312910) * exiv2: somehow this managed to slip through stagings: the package did not build with gcc5, due to missing includes (unistd.h). Only when cmake as buildsystem is used, then the variable would be correctly defined. Strange issue. But a switch to cmake building is submitted in SR#312666 and exiv2 based failures are confirmed to be fixed with this * p11-kit: the 'trust' binary (which is used to create the entire SSL Cert store) has issues to be started (reports ffi_prep_cif failed: 2). This error seems responsible for most (if not all) of the java build fails as well as the KDE3 related package failures. At this moment, there is no fix pending yet for this package. A bug report has been filed: https://bugzilla.opensuse.org/show_bug.cgi?id=935510 ) so, two out of three major issues are addressed; anybody finding time/energy/wisdom to look into p11-kit could possibly move that one forward (I strongly recommend to enable make check in p11-kit: it fails with the same errors at this moment). Cheers, and keep up the good work! Dominique -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Sun, Jun 21, 2015 at 7:37 AM, <dimstar@opensuse.org> wrote:
* libusb-compat: a patch that attempted to fix closed source softare has some negative effects. An SR, disabling the patch for now, is pending against the devel prj (SR#312910)
In the meanwhile I have reduced the number of packages that (build)require this library. some components have already migrated to libusb1 since the rpm specs were written, others do not use it and so on.. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Sun, Jun 21, 2015 at 1:31 PM, Cristian Rodríguez <crrodriguez@opensuse.org> wrote:
On Sun, Jun 21, 2015 at 7:37 AM, <dimstar@opensuse.org> wrote:
* libusb-compat: a patch that attempted to fix closed source softare has some negative effects. An SR, disabling the patch for now, is pending against the devel prj (SR#312910)
In the meanwhile I have reduced the number of packages that (build)require this library. some components have already migrated to libusb1 since the rpm specs were written, others do not use it and so on..
From the "important things" only plasma 5 (!!) and gpg still uses the old libusb..the other packages are secondary concerns. -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Sunday 05 July 2015 16:05:31 Cristian Rodríguez wrote:
On Sun, Jun 21, 2015 at 1:31 PM, Cristian Rodríguez
<crrodriguez@opensuse.org> wrote:
On Sun, Jun 21, 2015 at 7:37 AM, <dimstar@opensuse.org> wrote:
* libusb-compat: a patch that attempted to fix closed source softare has some negative effects. An SR, disabling the patch for now, is pending against the devel prj (SR#312910)
In the meanwhile I have reduced the number of packages that (build)require this library. some components have already migrated to libusb1 since the rpm specs were written, others do not use it and so on..
From the "important things" only plasma 5 (!!) and gpg still uses the old libusb..the other packages are secondary concerns.
Plasma uses it for the mouse configuration panel. It is used for configuration of older Logitech mice, the code is probably derived from LoMoCo. 1. It is used to configure some "advanced" functionality, e.g. sensor resolution, but nothing basic 2. It only works for quite old mice (one of the newest, MX1000 cordless laser, is from 2004) [1] 3. It does not work out of the box anyway, due to unsufficient permissions for direct device access: terminal output: kcmshell5: Error opening usbfs file: Keine Berechtigung strace output: [pid 13231] open("/dev/bus/usb/002/010", O_RDWR) = -1 EACCES (Permission denied) "You have a Logitech Mouse connected, and libusb was found at compile time, but it was not possible to access this mouse. This is probably caused by a permissions problem - you should consult the manual on how to fix this." So, lets just disable it. Kind regards, Stefan [1] http://quickgit.kde.org/?p=plasma-desktop.git&a=blob&f=kcms%2Finput%2Fmouse.cpp -- Stefan Brüns / Bergstraße 21 / 52062 Aachen home: +49 241 53809034 mobile: +49 151 50412019 work: +49 2405 49936-424 -- To unsubscribe, e-mail: opensuse-factory+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse-factory+owner@opensuse.org
On Monday, July 06, 2015 01:41:16 AM Stefan Bruens wrote:
On Sunday 05 July 2015 16:05:31 Cristian Rodríguez wrote:
On Sun, Jun 21, 2015 at 1:31 PM, Cristian Rodríguez
<crrodriguez@opensuse.org> wrote:
On Sun, Jun 21, 2015 at 7:37 AM, <dimstar@opensuse.org> wrote:
* libusb-compat: a patch that attempted to fix closed source softare has some negative effects. An SR, disabling the patch for now, is pending against the devel prj (SR#312910)
In the meanwhile I have reduced the number of packages that (build)require this library. some components have already migrated to libusb1 since the rpm specs were written, others do not use it and so on..
From the "important things" only plasma 5 (!!) and gpg still uses the old libusb..the other packages are secondary concerns.
Plasma uses it for the mouse configuration panel. It is used for configuration of older Logitech mice, the code is probably derived from LoMoCo.
1. It is used to configure some "advanced" functionality, e.g. sensor resolution, but nothing basic 2. It only works for quite old mice (one of the newest, MX1000 cordless laser, is from 2004) [1] 3. It does not work out of the box anyway, due to unsufficient permissions for direct device access:
terminal output: kcmshell5: Error opening usbfs file: Keine Berechtigung strace output: [pid 13231] open("/dev/bus/usb/002/010", O_RDWR) = -1 EACCES (Permission denied)
"You have a Logitech Mouse connected, and libusb was found at compile time, but it was not possible to access this mouse. This is probably caused by a permissions problem - you should consult the manual on how to fix this."
So, lets just disable it.
Follow up, after Stefan's submit request i've sent 2 patches upstream to remove logitech/libusb related code, and allow rest of the kcm to build. So with plasma 5.4, this code is no more. Cheers, Hrvoje
Kind regards,
Stefan
[1] http://quickgit.kde.org/?p=plasma-desktop.git&a=blob&f=kcms%2Finput%2Fmous e.cpp
participants (7)
-
Cristian Rodríguez -
dimstar＠opensuse.org -
Dominique Leuenberger a.k.a. Dimstar
-
Ilya Chernykh -
Stanislav Brabec -
Stefan Bruens -
šumski