Bug#841098: not fixed in rygel 0.32.1-1
On Mon, 26 Dec 2016, Andreas Henriksson wrote: > If you can't find anyone else in pkg-gnome that wants to cater to > your needs, then please consider rygel package orphaned. > In that case you're free to QA upload, please set the maintainer > field accordingly if so. I don't want to be involved anymore. Michael, I asked the MIA team and they told me that this statement from Andreas does not count as "orphaning the package". Would be possible for you to take over the package, since you are in the GNOME team, or, alternatively, tell Andreas to orphan the package following the established procedure? Thanks.
Bug#841098: not fixed in rygel 0.32.1-1
Hello Santiago Vila, If you can't find anyone else in pkg-gnome that wants to cater to your needs, then please consider rygel package orphaned. In that case you're free to QA upload, please set the maintainer field accordingly if so. I don't want to be involved anymore. Regards, Andreas Henriksson
Bug#841098: not fixed in rygel 0.32.1-1
On Sun, 13 Nov 2016, Andreas Henriksson wrote: > If you really think that a test-suite detecting an issue should > be solved not by fixing the issue (or fixing a potential flakyness > in the test) but rather by disabling the test No, I don't think at all that a test-suite detecting an issue should be solved by disabling the test. If that's what you understood, I didn't explain well enough. The patch was *only* to make the package build again reliably. If you don't disable the test and instead do nothing, then you still have the FTBFS bug, which is still serious and RC, no matter the joke severity of the day to which you want to downgrade it. > you should > be sending a patch against debhelper to remove dh_auto_test > from the sequiencer in dh. I could well suggest the exact opposite: If you think that a package having a test suite which is wrongly designed should be fixed by trying and trying again until it builds, you should be sending a patch against dpkg-dev so that dpkg-buildpackage builds the package automatically several times until it builds ok. Thanks.
Bug#841098: not fixed in rygel 0.32.1-1
Andreas Henriksson wrote: > Claiming this is RC is a bit over the top though. Not at all unless having two CPUs becomes part of build-essential definition, something, which, AFAIK, has not happened yet. > If you're only out > to ride the policy train without actually wanting to debug it yourself It's not that I don't want to debug it myself, it's just that I simply can't debug every bug I report, not to mention I would probably not have the required skills to debug this one. But not wanting to debug a bug should not be an excuse to accept the bug being just downgraded and forgotten. > then I'll be forced to simply disable the testsuite. Please do so (but only the test which fails) because otherwise it will be me who will be forced to appeal to the technical committee. As a last test, I've built this package 100 times today on a lot of *different* autobuilders (all of them having a single CPU). I've put the results here for you to see: https://people.debian.org/~sanvila/rygel/ Summary: It failed 95 times out of 100 times. This is a failure rate of 95% on single CPU machines. Unless having two CPUs has suddenly become part of the build-essential definition, I don't see how can you dispute that this is serious and RC. The fact that you downgraded this to *wishlist* (not important, not normal, not even minor, but wishlist) is really what is over the top here. > In my opinion > the testsuite exists to raise the quality of what we produce and > to *save* us time. Sure, see how much time we have saved discussing about this. A test only saves time when you do something when it fails. If you do nothing when it fails, it serves no purpose and does not improve the quality in any way. Thanks.
Bug#841098: not fixed in rygel 0.32.1-1
Control: tags -1 - patch Hello Santiago Vila. Thanks for trying to contribute, but you're patch is fundamentally flawed. On Fri, Oct 21, 2016 at 02:07:28PM +0200, Santiago Vila wrote: > tags 841098 + patch > thanks > > Here is a patch. After applying it the package builds fine and without > failures. > If you really think that a test-suite detecting an issue should be solved not by fixing the issue (or fixing a potential flakyness in the test) but rather by disabling the test you should be sending a patch against debhelper to remove dh_auto_test from the sequiencer in dh. Also when you're patching upstream sources I'd appreciate if you send your patches via upstream QA (and ask me to cherry-pick them if urgent once they've been applied in upstream git). Regards, Andreas Henriksson
Bug#841098: closed by Andreas Henriksson <andr...@fatal.se> (Re: Bug#841098: not fixed in rygel 0.32.1-1)
Control: severity -1 wishlist Hello Guillem Jover. Thanks for your input on this bug report... Unfortunately I missed it tough, not sure why it didn't reach me as you seem to have included me in the recipients. I also just read your reasoning in https://lists.debian.org/debian-devel/2016/11/msg00538.html that if you don't intend to work on something (which if I understood your mail correctly you don't intend here), or maybe your point was that the maintainer doesn't intend to work on it (and I don't, but I'm just an uploader OTOH the Debian GNOME team mostly relies on me taking care of rygel) then the correct severity is wishlist. I'm thus setting that severity on this bug report despite not ever having seen something in policy to support your reasoning, but I trust you On Sun, Oct 23, 2016 at 04:35:31PM +0200, Guillem Jover wrote: > Control: reopen -1 > > Hi! > > On Thu, 2016-10-20 at 20:15:06 +, Debian Bug Tracking System wrote: > > Date: Thu, 20 Oct 2016 22:14:11 +0200 > > From: Andreas Henriksson <andr...@fatal.se> > > To: Santiago Vila <sanv...@unex.es> > > Cc: 841098-d...@bugs.debian.org > > Subject: Re: Bug#841098: not fixed in rygel 0.32.1-1 > > Message-ID: <20161020201411.ga27...@fatal.se> > > User-Agent: Mutt/1.5.23 (2014-03-12) > > X-Spam-Status: No, score=-8.3 required=4.0 tests=BAYES_00,HAS_BUG_NUMBER, > > RP_MATCHES_RCVD,VERSION autolearn=ham autolearn_force=no > > version=3.4.0-bugs.debian.org_2005_01_02 > > > The above mentioned upload was done as a source-only upload. > > It built successfully not only in my local environment but also on *all* > > architectures (and thus on atleast one buildd for each architecture). > > > This is obviously a problem in your environment. > > As Santiago has shown this has also happened on the buildds, very > recent example being: > > > <https://buildd.debian.org/status/fetch.php?pkg=rygel=ppc64el=0.32.0-2=1475601506> > > <https://buildd.debian.org/status/fetch.php?pkg=rygel=s390x=0.32.0-1=1474571422> > > And he can still reproduce it. If this is, say, a timing issue, then it > might not always manifest itself. It does not mean it's not present. A > segfault in the buildds is a clear indication this is not an imagined > situation. > > > The debian bug tracking system is not a support forum. If you need > > personal support, then you need to negotiate a support contract first. > > The debian bug tracking system is not the place for this. > > I'm sorry to say but I feel this comment is inappropriate. Santiago is > doing distro-wide QA, I'd even go as far as to hazard he does not even > use this package at all! Me neither but I still have to fix it or what's your point? If QA in Debian is defined as "pointing out stuff noone cares about and making a fuzz about it" then I think we should simply stop doing QA and instead focus on real bugs and to limit the scope even more focus on things that actually affects real users. Patches still welcome for this issue though! Regards, Andreas Henriksson
Bug#841098: not fixed in rygel 0.32.1-1
On Mon, Oct 24, 2016 at 01:45:24PM +0200, Andreas Henriksson wrote: > If you're strongly attached to (cluttering up my maintainer view and) > tracking every issue in the debian bug tracking system (I'd instead > recommend you use the upstream bug tracker), then I'd welcome > a bug report with severity minor. Minor is in my opinion the correct > severity for such an issue. I'm only attached to ensuring that Debian packages build ok and without failures (release policy), so I have to decline your suggestion to use the upstream bug tracker, as that will not make the Debian package to build ok 100% of the time again. If you think this is not RC, you can always ask for permission to use sarge-ignore here (you could ask in the same report if it's ok for them to disable the test). IMO, disabling the test would be just a way to follow release policy, which says "packages must build without failures". If you plan to downgrade instead, please consider these two bugs as well: https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=839444 https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=811068 The three bugs are FTBFS which happen randomly. The package usually builds fine in the official autobuilders (but not always), and the FTBFS problem could be avoided trivially by disabling one or more tests. It would be really strange to downgrade one of those bugs and not the others. Before doing so, please try to reach a consensus inside the GNOME team about what should be done with all those bugs. Thanks.
Bug#841098: not fixed in rygel 0.32.1-1
Hello Santiago Vila. Thanks for the backtrace, it's helpful. On Fri, Oct 21, 2016 at 01:16:20PM +0200, Santiago Vila wrote: > The package FTBFS (sometimes) because rygel-media-engine-test segfaults. > > Here is a backtrace: > > (gdb) bt > #0 0x7f1a1a4b3fdf in raise () from /lib/x86_64-linux-gnu/libc.so.6 > #1 0x7f1a1a4b540a in abort () from /lib/x86_64-linux-gnu/libc.so.6 > #2 0x7f1a1af8a4f5 in g_assertion_message () from > /lib/x86_64-linux-gnu/libglib-2.0.so.0 This is not a segfault though, but an assertion failure. > #3 0x7f1a1af8a58a in g_assertion_message_expr () from > /lib/x86_64-linux-gnu/libglib-2.0.so.0 > #4 0x55597c5578bd in __lambda19_ (_data3_=) at > rygel-media-engine-test.c:2035 [...] ^^^ hint on where the failing assertion can be found. (Please note that the source is vala so you'll need to translate this to the applicable lambda in the vala source.) > #5 ___lambda19__gsource_func (self=0x7f19f4001410) at > rygel-media-engine-test.c:2047 > #6 0x7f1a1af6368a in g_main_context_dispatch () from > /lib/x86_64-linux-gnu/libglib-2.0.so.0 > #7 0x7f1a1af63a40 in ?? () from /lib/x86_64-linux-gnu/libglib-2.0.so.0 > #8 0x7f1a1af63d62 in g_main_loop_run () from > /lib/x86_64-linux-gnu/libglib-2.0.so.0 > #9 0x55597c559688 in rygel_data_source_test_test_stop_start > (error=0x7ffe67fae238, self=0x55597df7e4e0) > at rygel-media-engine-test.c:2116 > #10 rygel_data_source_test_run (self=0x55597df7e4e0) at > rygel-media-engine-test.c:2717 > #11 0x55597c55a310 in rygel_data_source_test_main (args=, > args_length1=) > at rygel-media-engine-test.c:2996 > #12 0x7f1a1a4a12b1 in __libc_start_main () from > /lib/x86_64-linux-gnu/libc.so.6 > #13 0x55597c556f4a in _start () > > It could be, of course, that my autobuilder is misconfigured, but the > same segfault happened several times in buildd.debian.org in the past. You're more than welcome to dive deep into this issue. Patches are always welcome. Claiming this is RC is a bit over the top though. If you're only out to ride the policy train without actually wanting to debug it yourself then I'll be forced to simply disable the testsuite. In my opinion the testsuite exists to raise the quality of what we produce and to *save* us time. If it becomes to time-consuming to maintain I'll do without it. Disabling failing tests-suites is something I know from personal experience that several release team members have a strong opinion against. Disabling it for a extremely rare occational failure is something I imagine they would be even more opposed to. If you're strongly attached to (cluttering up my maintainer view and) tracking every issue in the debian bug tracking system (I'd instead recommend you use the upstream bug tracker), then I'd welcome a bug report with severity minor. Minor is in my opinion the correct severity for such an issue. Regards, Andreas Henriksson
Bug#841098: closed by Andreas Henriksson <andr...@fatal.se> (Re: Bug#841098: not fixed in rygel 0.32.1-1)
Control: reopen -1 Hi! On Thu, 2016-10-20 at 20:15:06 +, Debian Bug Tracking System wrote: > Date: Thu, 20 Oct 2016 22:14:11 +0200 > From: Andreas Henriksson <andr...@fatal.se> > To: Santiago Vila <sanv...@unex.es> > Cc: 841098-d...@bugs.debian.org > Subject: Re: Bug#841098: not fixed in rygel 0.32.1-1 > Message-ID: <20161020201411.ga27...@fatal.se> > User-Agent: Mutt/1.5.23 (2014-03-12) > X-Spam-Status: No, score=-8.3 required=4.0 tests=BAYES_00,HAS_BUG_NUMBER, > RP_MATCHES_RCVD,VERSION autolearn=ham autolearn_force=no > version=3.4.0-bugs.debian.org_2005_01_02 > The above mentioned upload was done as a source-only upload. > It built successfully not only in my local environment but also on *all* > architectures (and thus on atleast one buildd for each architecture). > This is obviously a problem in your environment. As Santiago has shown this has also happened on the buildds, very recent example being: <https://buildd.debian.org/status/fetch.php?pkg=rygel=ppc64el=0.32.0-2=1475601506> <https://buildd.debian.org/status/fetch.php?pkg=rygel=s390x=0.32.0-1=1474571422> And he can still reproduce it. If this is, say, a timing issue, then it might not always manifest itself. It does not mean it's not present. A segfault in the buildds is a clear indication this is not an imagined situation. > The debian bug tracking system is not a support forum. If you need > personal support, then you need to negotiate a support contract first. > The debian bug tracking system is not the place for this. I'm sorry to say but I feel this comment is inappropriate. Santiago is doing distro-wide QA, I'd even go as far as to hazard he does not even use this package at all! Thanks, Guillem
Bug#841098: not fixed in rygel 0.32.1-1
Sorry, there was a typo. Here is a fixed patch. Thanks.diff -Nru rygel-0.32.1/debian/changelog rygel-0.32.1/debian/changelog --- rygel-0.32.1/debian/changelog 2016-10-18 09:21:52.0 +0200 +++ rygel-0.32.1/debian/changelog 2016-10-21 13:49:26.0 +0200 @@ -1,3 +1,10 @@ +rygel (0.32.1-2) unstable; urgency=medium + + * Disable rygel-media-engine-test, as it fails randomly, making the +package to FTBFS (Closes: #841098). + + -- Andreas HenrikssonFri, 21 Oct 2016 13:49:26 +0200 + rygel (0.32.1-1) unstable; urgency=medium * Drop obsolete transitional package (Closes: #841098) diff -Nru rygel-0.32.1/debian/patches/001-disable-rygel-media-engine-test.patcb rygel-0.32.1/debian/patches/001-disable-rygel-media-engine-test.patcb --- rygel-0.32.1/debian/patches/001-disable-rygel-media-engine-test.patcb 1970-01-01 01:00:00.0 +0100 +++ rygel-0.32.1/debian/patches/001-disable-rygel-media-engine-test.patcb 2016-10-21 13:49:26.0 +0200 @@ -0,0 +1,17 @@ +From: Santiago Vila +Subject: Disable rygel-media-engine-test until we know why it segfaults +Bug-Debian: https://bugs.debian.org/841098 + +--- rygel-0.32.1.orig/tests/Makefile.am rygel-0.32.1/tests/Makefile.am +@@ -15,8 +15,8 @@ check_PROGRAMS = rygel-regression \ +rygel-user-config-test \ +rygel-searchable-container-test \ +rygel-object-creator-test \ +- rygel-database-test \ +- rygel-media-engine-test ++ rygel-database-test ++# rygel-media-engine-test + + TESTS = $(check_PROGRAMS) + diff -Nru rygel-0.32.1/debian/patches/series rygel-0.32.1/debian/patches/series --- rygel-0.32.1/debian/patches/series 1970-01-01 01:00:00.0 +0100 +++ rygel-0.32.1/debian/patches/series 2016-10-21 13:46:18.0 +0200 @@ -0,0 +1 @@ +001-disable-rygel-media-engine-test.patcb
Bug#841098: not fixed in rygel 0.32.1-1
tags 841098 + patch thanks Here is a patch. After applying it the package builds fine and without failures. Thanks.diff -Nru rygel-0.32.1/debian/changelog rygel-0.32.1/debian/changelog --- rygel-0.32.1/debian/changelog 2016-10-18 09:21:52.0 +0200 +++ rygel-0.32.1/debian/changelog 2016-10-21 13:49:26.0 +0200 @@ -1,3 +1,10 @@ +rygel (0.32.1-2) unstable; urgency=medium + + * Disable rygel-media-engine-test, as it fails randomly, making the +package to FTBFS (Closes: #841098). + + -- Andreas HenrikssonFri, 21 Oct 2016 13:49:26 +0200 + rygel (0.32.1-1) unstable; urgency=medium * Drop obsolete transitional package (Closes: #841098) diff -Nru rygel-0.32.1/debian/patches/001-disable-rygel-media-engine-test.patcb rygel-0.32.1/debian/patches/001-disable-rygel-media-engine-test.patcb --- rygel-0.32.1/debian/patches/001-disable-rygel-media-engine-test.patcb 1970-01-01 01:00:00.0 +0100 +++ rygel-0.32.1/debian/patches/001-disable-rygel-media-engine-test.patcb 2016-10-21 13:47:41.0 +0200 @@ -0,0 +1,17 @@ +From: Santiago Vila +Subject: Disable rygel-media-engine-test until we know what it segfaults +Bug-Debian: https://bugs.debian.org/841098 + +--- rygel-0.32.1.orig/tests/Makefile.am rygel-0.32.1/tests/Makefile.am +@@ -15,8 +15,8 @@ check_PROGRAMS = rygel-regression \ +rygel-user-config-test \ +rygel-searchable-container-test \ +rygel-object-creator-test \ +- rygel-database-test \ +- rygel-media-engine-test ++ rygel-database-test ++# rygel-media-engine-test + + TESTS = $(check_PROGRAMS) + diff -Nru rygel-0.32.1/debian/patches/series rygel-0.32.1/debian/patches/series --- rygel-0.32.1/debian/patches/series 1970-01-01 01:00:00.0 +0100 +++ rygel-0.32.1/debian/patches/series 2016-10-21 13:46:18.0 +0200 @@ -0,0 +1 @@ +001-disable-rygel-media-engine-test.patcb
Bug#841098: not fixed in rygel 0.32.1-1
The package FTBFS (sometimes) because rygel-media-engine-test segfaults. Here is a backtrace: (gdb) bt #0 0x7f1a1a4b3fdf in raise () from /lib/x86_64-linux-gnu/libc.so.6 #1 0x7f1a1a4b540a in abort () from /lib/x86_64-linux-gnu/libc.so.6 #2 0x7f1a1af8a4f5 in g_assertion_message () from /lib/x86_64-linux-gnu/libglib-2.0.so.0 #3 0x7f1a1af8a58a in g_assertion_message_expr () from /lib/x86_64-linux-gnu/libglib-2.0.so.0 #4 0x55597c5578bd in __lambda19_ (_data3_=) at rygel-media-engine-test.c:2035 #5 ___lambda19__gsource_func (self=0x7f19f4001410) at rygel-media-engine-test.c:2047 #6 0x7f1a1af6368a in g_main_context_dispatch () from /lib/x86_64-linux-gnu/libglib-2.0.so.0 #7 0x7f1a1af63a40 in ?? () from /lib/x86_64-linux-gnu/libglib-2.0.so.0 #8 0x7f1a1af63d62 in g_main_loop_run () from /lib/x86_64-linux-gnu/libglib-2.0.so.0 #9 0x55597c559688 in rygel_data_source_test_test_stop_start (error=0x7ffe67fae238, self=0x55597df7e4e0) at rygel-media-engine-test.c:2116 #10 rygel_data_source_test_run (self=0x55597df7e4e0) at rygel-media-engine-test.c:2717 #11 0x55597c55a310 in rygel_data_source_test_main (args=, args_length1=) at rygel-media-engine-test.c:2996 #12 0x7f1a1a4a12b1 in __libc_start_main () from /lib/x86_64-linux-gnu/libc.so.6 #13 0x55597c556f4a in _start () It could be, of course, that my autobuilder is misconfigured, but the same segfault happened several times in buildd.debian.org in the past. Thanks.
Bug#841098: not fixed in rygel 0.32.1-1
On Thu, Oct 20, 2016 at 10:14:11PM +0200, Andreas Henriksson wrote: > The above mentioned upload was done as a source-only upload. > It built successfully not only in my local environment but also on *all* > architectures (and thus on atleast one buildd for each architecture). > > This is obviously a problem in your environment. No, I think it's not, because it also happened in the official autobuilders several times. For example: https://buildd.debian.org/status/fetch.php?pkg=rygel=ppc64el=0.32.0-2=1475601506 Quote: PASS: rygel-searchable-container-test PASS: rygel-object-creator-test PASS: rygel-database-test ../build-aux/test-driver: line 107: 17241 Aborted "$@" > $log_file 2>&1 FAIL: rygel-media-engine-test PASS: rygel-playbin-renderer-test PASS: rygel-environment-test It's more like a bug which happens randomly (with low probability) and it's still unfixed, but definitely not a problem in my environment. Thanks.