Debian Bug report logs - #877168
transition: ldc

Reply or subscribe to this bug.

Toggle useless messages

Message #5 received at submit@bugs.debian.org (full text, mbox, reply):

From: Michael Biebl <biebl@debian.org>

To: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: nmu: gtk-d_3.6.5-2

Date: Fri, 29 Sep 2017 13:38:00 +0200

Package: release.debian.org
Severity: normal
User: release.debian.org@packages.debian.org
Usertags: binnmu

nmu gtk-d_3.6.5-2 . amd64 . unstable . -m "Rebuild against libphobos2-ldc74"

tilix was built with the latest version of ldc and as a result is linked
against libphobos2-ldc74. It also links against libvted-3-0, which was
not compiled with the newest ldc version yet, so still links against
libphobos2-ldc72. Apparently mixing libraries this way is not supported
and as a consequence tilix fails to start, see #876479:

$ tilix
Fatal Error while loading '/usr/lib/x86_64-linux-gnu/libdruntime-ldc.so.72':
	The module 'core.atomic' is already defined in '/usr/lib/x86_64-linux-gnu/libdruntime-ldc.so.74'.

Rebuilding gtk-d (which provides libvted-3-0) against libphobos2-ldc74
seems to fix the error.

CCed the D developers. It would be nice if such situations could be
avoided in the future. Maybe you have some ideas how to achieve that.

The situation is a mess:
https://packages.debian.org/sid/tilix
https://packages.debian.org/sid/libvted-3-0

There is no consistency wrt to which version of libphobos2-ldc the
binaries are linked against.

Regards,
Michael

-- System Information:
Debian Release: buster/sid
  APT prefers unstable
  APT policy: (500, 'unstable'), (200, 'experimental')
Architecture: amd64 (x86_64)
Foreign Architectures: i386

Kernel: Linux 4.12.0-2-amd64 (SMP w/4 CPU cores)
Locale: LANG=de_DE.utf8, LC_CTYPE=de_DE.utf8 (charmap=UTF-8), LANGUAGE=de_DE.utf8 (charmap=UTF-8)
Shell: /bin/sh linked to /bin/dash
Init: systemd (via /run/systemd/system)

Message #10 received at 877168@bugs.debian.org (full text, mbox, reply):

From: Matthias Klumpp <mak@debian.org>

To: Michael Biebl <biebl@debian.org>, 877168@bugs.debian.org

Cc: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: Re: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Fri, 29 Sep 2017 13:45:48 +0200

2017-09-29 13:38 GMT+02:00 Michael Biebl <biebl@debian.org>:
> Package: release.debian.org
> Severity: normal
> User: release.debian.org@packages.debian.org
> Usertags: binnmu
>
> nmu gtk-d_3.6.5-2 . amd64 . unstable . -m "Rebuild against libphobos2-ldc74"
>
> tilix was built with the latest version of ldc and as a result is linked
> against libphobos2-ldc74. It also links against libvted-3-0, which was
> not compiled with the newest ldc version yet, so still links against
> libphobos2-ldc72. Apparently mixing libraries this way is not supported
> and as a consequence tilix fails to start, see #876479:
>
> $ tilix
> Fatal Error while loading '/usr/lib/x86_64-linux-gnu/libdruntime-ldc.so.72':
>         The module 'core.atomic' is already defined in '/usr/lib/x86_64-linux-gnu/libdruntime-ldc.so.74'.
>
> Rebuilding gtk-d (which provides libvted-3-0) against libphobos2-ldc74
> seems to fix the error.
>
> CCed the D developers. It would be nice if such situations could be
> avoided in the future. Maybe you have some ideas how to achieve that.
>
> The situation is a mess:
> https://packages.debian.org/sid/tilix
> https://packages.debian.org/sid/libvted-3-0
>
> There is no consistency wrt to which version of libphobos2-ldc the
> binaries are linked against.

Agreed. Ideally we'd need to coordinate binNMUs on new ABI versions
better, so thinks are rebuilt in order, and not randomly. Also, LDC
needs to migrate to testing faster (currently blocked on rewriting the
d/copyright file).
Maybe we can also add a breaks relationship to the newer libphobos
versions, so mixing runtimes in software is impossible.

Thank you for filing the binNMU request!

Cheers,
    Matthias

-- 
I welcome VSRE emails. See http://vsre.info/

Message #15 received at submit@bugs.debian.org (full text, mbox, reply):

From: Emilio Pozuelo Monfort <pochu@debian.org>

To: Matthias Klumpp <mak@debian.org>, 877168@bugs.debian.org, Michael Biebl <biebl@debian.org>

Cc: Debian Bug Tracking System <submit@bugs.debian.org>

Subject: Re: Bug#877168: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Sat, 30 Sep 2017 11:30:27 +0200

On 29/09/17 13:45, Matthias Klumpp wrote:
> 2017-09-29 13:38 GMT+02:00 Michael Biebl <biebl@debian.org>:
>> Package: release.debian.org
>> Severity: normal
>> User: release.debian.org@packages.debian.org
>> Usertags: binnmu
>>
>> nmu gtk-d_3.6.5-2 . amd64 . unstable . -m "Rebuild against libphobos2-ldc74"

I tried this a while ago against -ldc73, but it failed on amd64, see:

https://buildd.debian.org/status/package.php?p=gtk-d

Is this going to work now? If so, I should rebuild this on all architectures,
not just amd64, right?

Also, what about the other rdepends? Some of them failed to build too in the
past, so I eventually gave up, waiting for a bug like this to come up :)

https://release.debian.org/transitions/html/auto-ldc.html

> Agreed. Ideally we'd need to coordinate binNMUs on new ABI versions
> better, so thinks are rebuilt in order, and not randomly.

Some coordination when things need special treatment is always good :)

Let me know what is needed here and I'll help if I can.

Cheers,
Emilio

Message #25 received at 877168@bugs.debian.org (full text, mbox, reply):

From: Michael Biebl <biebl@debian.org>

To: Emilio Pozuelo Monfort <pochu@debian.org>, 877168@bugs.debian.org, Matthias Klumpp <mak@debian.org>

Subject: Re: Bug#877168: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Sat, 30 Sep 2017 12:46:59 +0200

[Message part 1 (text/plain, inline)]

Am 30.09.2017 um 11:30 schrieb Emilio Pozuelo Monfort:

> Is this going to work now? If so, I should rebuild this on all architectures,
> not just amd64, right?

I did a successful test-build in a sid amd64 chroot. If that means it
will work on the buildds, I dunno.

As for rebuilding on all architectures: If you do that, you should
probably also rebuild tilix on all architectures which do not link
against libphobos2-ldc74 (like say ppc64el or armhf)




-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?

[signature.asc (application/pgp-signature, attachment)]

Message #30 received at submit@bugs.debian.org (full text, mbox, reply):

From: Matthias Klumpp <mak@debian.org>

To: Emilio Pozuelo Monfort <pochu@debian.org>

Cc: 877168@bugs.debian.org, Michael Biebl <biebl@debian.org>, Debian Bug Tracking System <submit@bugs.debian.org>

Subject: Re: Bug#877168: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Sat, 30 Sep 2017 15:52:23 +0200

2017-09-30 11:30 GMT+02:00 Emilio Pozuelo Monfort <pochu@debian.org>:
> On 29/09/17 13:45, Matthias Klumpp wrote:
>> 2017-09-29 13:38 GMT+02:00 Michael Biebl <biebl@debian.org>:
>>> Package: release.debian.org
>>> Severity: normal
>>> User: release.debian.org@packages.debian.org
>>> Usertags: binnmu
>>>
>>> nmu gtk-d_3.6.5-2 . amd64 . unstable . -m "Rebuild against libphobos2-ldc74"
>
> I tried this a while ago against -ldc73, but it failed on amd64, see:
>
> https://buildd.debian.org/status/package.php?p=gtk-d
>
> Is this going to work now? If so, I should rebuild this on all architectures,
> not just amd64, right?

Yes, it should work now. I got a report of sambamba failing to build,
but that could have been an artifact of mixing runtimes again, so I
think rebuilding everything in order is a very good idea at time.
LDC was hit by a nasty LLVM bug, which we "resolved" by rebuilding LDC
against LLVM 5.0 (instead of 4.0) - currently, all builds I do on
amd64 with ldc 1:1.4.0-2 build & run without any issue.

> Also, what about the other rdepends? Some of them failed to build too in the
> past, so I eventually gave up, waiting for a bug like this to come up :)
>
> https://release.debian.org/transitions/html/auto-ldc.html

That should work now as well - the important thing is that packages
are rebuilt in the right order, so no package accidentally picks up
symbols from an older runtime version.
In order to do that, could you augment the LDC Ben tracker to consider
anything depending on libphobos2-ldc72 to be bad as well? I think
there are a few packages missing on the tracker (sambamba, for
example).

>> Agreed. Ideally we'd need to coordinate binNMUs on new ABI versions
>> better, so thinks are rebuilt in order, and not randomly.
>
> Some coordination when things need special treatment is always good :)
>
> Let me know what is needed here and I'll help if I can.

See above :-) Thank you for your work and sorry for the mess, this
should have been a pretty easy transition - we didn't anticipate the
whole world breaking with a weird crash in LLVM :P

I hope it works now (but the only way to know for sure is to rebuild
on all architectures - amd64 is fine, at least)

Cheers,
    Matthias

-- 
I welcome VSRE emails. See http://vsre.info/

Message #40 received at submit@bugs.debian.org (full text, mbox, reply):

From: Matthias Klumpp <mak@debian.org>

To: Emilio Pozuelo Monfort <pochu@debian.org>

Cc: 877168@bugs.debian.org, Michael Biebl <biebl@debian.org>, Debian Bug Tracking System <submit@bugs.debian.org>

Subject: Re: Bug#877168: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Sun, 1 Oct 2017 02:32:52 +0200

2017-09-30 15:52 GMT+02:00 Matthias Klumpp <mak@debian.org>:
> 2017-09-30 11:30 GMT+02:00 Emilio Pozuelo Monfort <pochu@debian.org>:
>> [...]
>>> Agreed. Ideally we'd need to coordinate binNMUs on new ABI versions
>>> better, so thinks are rebuilt in order, and not randomly.
>>
>> Some coordination when things need special treatment is always good :)
>>
>> Let me know what is needed here and I'll help if I can.

What about this tracker, to really catch all the LDC stuff?

title = "ldc-phobos";
is_affected = .build-depends ~ /ldc/;
is_good = .depends ~ /\b(libphobos2\-ldc74)\b/;
is_bad = (.depends ~ /\b(libphobos2\-ldc73)\b/) | (.depends ~
/\b(libphobos2\-ldc72)\b/);
notes = "";

I couldn't test it, so maybe there are issues.
Thanks!
    Matthias

-- 
I welcome VSRE emails. See http://vsre.info/

Message #50 received at submit@bugs.debian.org (full text, mbox, reply):

From: Michael Biebl <biebl@debian.org>

To: Matthias Klumpp <mak@debian.org>, Emilio Pozuelo Monfort <pochu@debian.org>

Cc: 877168@bugs.debian.org, Debian Bug Tracking System <submit@bugs.debian.org>

Subject: Re: Bug#877168: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Sun, 1 Oct 2017 14:06:07 +0200

[Message part 1 (text/plain, inline)]

Hi Matthias,

I see that you made a sourceful upload of gtk-d, so I thought that this
binNMU request is now moot.
Unfortunately, tilix still doesn't want to start:

The error is a different one now though:
$ tilix
tilix: symbol lookup error: tilix: undefined symbol:
_D3gio9DBusProxy9DBusProxy9__mixin3622getAsyncInitableStructMFbZPS4gtkc8giotypes14GAsyncInitable


-- 
Why is it that all of the instruments seeking intelligent life in the
universe are pointed away from Earth?

[signature.asc (application/pgp-signature, attachment)]

Message #60 received at submit@bugs.debian.org (full text, mbox, reply):

From: Matthias Klumpp <mak@debian.org>

To: Michael Biebl <biebl@debian.org>

Cc: Emilio Pozuelo Monfort <pochu@debian.org>, 877168@bugs.debian.org, Debian Bug Tracking System <submit@bugs.debian.org>

Subject: Re: Bug#877168: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Sun, 1 Oct 2017 16:29:01 +0200

2017-10-01 14:06 GMT+02:00 Michael Biebl <biebl@debian.org>:
> Hi Matthias,
>
> I see that you made a sourceful upload of gtk-d, so I thought that this
> binNMU request is now moot.
> Unfortunately, tilix still doesn't want to start:
>
> The error is a different one now though:
> $ tilix
> tilix: symbol lookup error: tilix: undefined symbol:
> _D3gio9DBusProxy9DBusProxy9__mixin3622getAsyncInitableStructMFbZPS4gtkc8giotypes14GAsyncInitable

Hmm, weird, it looks like GtkD broke ABI then - so now we need to
rebuild Tilix again as well. I think the only long-term solution here,
while GtkD and D don't have an ABI stability promise, is to bump the
soname of GtkD libraries constantly.
Looks like others are also having fun with this:
https://github.com/gnunn1/tilix/issues/1064

I will make a new Tilix upload, solving this issue, and starting with
the next GtkD release, I'll just bump the SOVERSION every time (maybe
I can also get upstream to do that).

Cheers,
    Matthias


-- 
I welcome VSRE emails. See http://vsre.info/

Message #72 received at 877168@bugs.debian.org (full text, mbox, reply):

From: Emilio Pozuelo Monfort <pochu@debian.org>

To: Matthias Klumpp <mak@debian.org>, 877168@bugs.debian.org, Michael Biebl <biebl@debian.org>

Subject: Re: Bug#877168: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Tue, 3 Oct 2017 19:30:55 +0200

Control: forwarded -1 https://release.debian.org/transitions/html/ldc.html
Control: tags -1 confirmed

On 01/10/17 02:32, Matthias Klumpp wrote:
> 2017-09-30 15:52 GMT+02:00 Matthias Klumpp <mak@debian.org>:
>> 2017-09-30 11:30 GMT+02:00 Emilio Pozuelo Monfort <pochu@debian.org>:
>>> [...]
>>>> Agreed. Ideally we'd need to coordinate binNMUs on new ABI versions
>>>> better, so thinks are rebuilt in order, and not randomly.
>>>
>>> Some coordination when things need special treatment is always good :)
>>>
>>> Let me know what is needed here and I'll help if I can.
> 
> What about this tracker, to really catch all the LDC stuff?
> 
> title = "ldc-phobos";
> is_affected = .build-depends ~ /ldc/;
> is_good = .depends ~ /\b(libphobos2\-ldc74)\b/;
> is_bad = (.depends ~ /\b(libphobos2\-ldc73)\b/) | (.depends ~
> /\b(libphobos2\-ldc72)\b/);
> notes = "";

I went for a slightly different one. I'll schedule the binNMUs later.

Cheers,
Emilio

Message #81 received at 877168@bugs.debian.org (full text, mbox, reply):

From: Emilio Pozuelo Monfort <pochu@debian.org>

To: Matthias Klumpp <mak@debian.org>, 877168@bugs.debian.org, Michael Biebl <biebl@debian.org>

Subject: Re: Bug#877168: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Wed, 4 Oct 2017 09:39:22 +0200

On 03/10/17 19:30, Emilio Pozuelo Monfort wrote:
> Control: forwarded -1 https://release.debian.org/transitions/html/ldc.html
> Control: tags -1 confirmed
> 
> On 01/10/17 02:32, Matthias Klumpp wrote:
>> 2017-09-30 15:52 GMT+02:00 Matthias Klumpp <mak@debian.org>:
>>> 2017-09-30 11:30 GMT+02:00 Emilio Pozuelo Monfort <pochu@debian.org>:
>>>> [...]
>>>>> Agreed. Ideally we'd need to coordinate binNMUs on new ABI versions
>>>>> better, so thinks are rebuilt in order, and not randomly.
>>>>
>>>> Some coordination when things need special treatment is always good :)
>>>>
>>>> Let me know what is needed here and I'll help if I can.
>>
>> What about this tracker, to really catch all the LDC stuff?
>>
>> title = "ldc-phobos";
>> is_affected = .build-depends ~ /ldc/;
>> is_good = .depends ~ /\b(libphobos2\-ldc74)\b/;
>> is_bad = (.depends ~ /\b(libphobos2\-ldc73)\b/) | (.depends ~
>> /\b(libphobos2\-ldc72)\b/);
>> notes = "";
> 
> I went for a slightly different one. I'll schedule the binNMUs later.

I did that, and some things built (libundead, libbiod) but others failed
(gir-to-d/ppc64el, sambamba).

Cheers,
Emilio

Message #86 received at 877168@bugs.debian.org (full text, mbox, reply):

From: Matthias Klumpp <mak@debian.org>

To: Emilio Pozuelo Monfort <pochu@debian.org>

Cc: 877168@bugs.debian.org, Michael Biebl <biebl@debian.org>

Subject: Re: Bug#877168: [Pkg-d-devel] Bug#877168: nmu: gtk-d_3.6.5-2

Date: Wed, 4 Oct 2017 13:36:25 +0200

2017-10-04 9:39 GMT+02:00 Emilio Pozuelo Monfort <pochu@debian.org>:
> [...]
> I did that, and some things built (libundead, libbiod) but others failed
> (gir-to-d/ppc64el, sambamba).

Thank you!
Both issues are reported upstream:
ppc64el: https://github.com/ldc-developers/ldc/issues/2356
sambamba assert: https://github.com/ldc-developers/ldc/issues/2357

I will report bugs on Debian as well later (but not make them RC
immediately, so the current LDC can migrate to testing and replace the
totally broken version that's currently in testing).
Both issues are LDC's fault, and I hope upstream can resolve them soonish.

Cheers,
    Matthias

-- 
I welcome VSRE emails. See http://vsre.info/

Message #91 received at 877168@bugs.debian.org (full text, mbox, reply):

From: Matthias Klumpp <mak@debian.org>

To: Emilio Pozuelo Monfort <pochu@debian.org>

Cc: 877168@bugs.debian.org, Michael Biebl <biebl@debian.org>

Subject: Re: Bug#877168: transition: ldc

Date: Tue, 7 Nov 2017 02:02:30 +0100

Hi!

2017-10-04 13:36 GMT+02:00 Matthias Klumpp <mak@debian.org>:
> 2017-10-04 9:39 GMT+02:00 Emilio Pozuelo Monfort <pochu@debian.org>:
>> [...]
> Thank you!
> Both issues are reported upstream:
> ppc64el: https://github.com/ldc-developers/ldc/issues/2356
> sambamba assert: https://github.com/ldc-developers/ldc/issues/2357

Just a quick heads up: There is a new version of LDC soon in unstable
that will fix at least the Sambamba issue, and maybe (hopefully?) also
the ppc64el issue.
This will restart this transition though - unstable ABIs are a lot of fun...

Cheers,
    Matthias

-- 
I welcome VSRE emails. See http://vsre.info/

Message #96 received at 877168@bugs.debian.org (full text, mbox, reply):

From: Gianfranco Costamagna <locutusofborg@debian.org>

To: 877168@bugs.debian.org

Subject: Re: Bug#877168: transition: ldc

Date: Tue, 7 Nov 2017 20:24:41 +0100

[Message part 1 (text/plain, inline)]

On Tue, 7 Nov 2017 02:02:30 +0100 Matthias Klumpp <mak@debian.org> wrote:
> Hi!
> 
> 2017-10-04 13:36 GMT+02:00 Matthias Klumpp <mak@debian.org>:
> > 2017-10-04 9:39 GMT+02:00 Emilio Pozuelo Monfort <pochu@debian.org>:
> >> [...]
> > Thank you!
> > Both issues are reported upstream:
> > ppc64el: https://github.com/ldc-developers/ldc/issues/2356
> > sambamba assert: https://github.com/ldc-developers/ldc/issues/2357
> 
> Just a quick heads up: There is a new version of LDC soon in unstable
> that will fix at least the Sambamba issue, and maybe (hopefully?) also
> the ppc64el issue.
> This will restart this transition though - unstable ABIs are a lot of fun...
> 

please update the tracker 74 -> 75

thanks

G.

[signature.asc (application/pgp-signature, attachment)]

Send a report that this bug log contains spam.