[Debian-med-packaging] New Sight version

Tue Jun 28 09:40:18 BST 2022

Hello Etienne,

I have started working on the issues you pointed out.

The build failure with gcc12 is fixed (I was very lucky to guess it, but 
I am really not sure to understand why it failed in the first place...).

I've addressed most lintian warnings (not info yet...), but I struggle a 
bit with the last one, about package sonames.

I am not sure to understand how I should fix it. I use a semantic 
version number for Sight. Usually we try to only break binary 
compatibility with major version (19, 20, 21). Lately we also broke 
binary compatibility with minor versions, but this should not happen. 
Whatever, currently in CMake we define this property to set the library 
so version.

set_target_properties(
${SIGHT_TARGET}PROPERTIESVERSION${PROJECT_VERSION}SOVERSION${PROJECT_VERSION_MAJOR}.${PROJECT_VERSION_MINOR}
)
So as far I understand with the snippet given on the lintian warning, it 
would expects the package to be named libsight22.0... So two questions:

 1. We do not have a library libsight.so but many libsight_name.so. Is
    this a problem for lintian or would I need to split into multiple
    packages ?
 2. Is the solution to rename the package to libsight22.0 ? Or maybe if
    I manage to be a bit stricter with development rules, stick to
    libsight22 ? If yes, does this mean on every major upgrade, I need
    to change the name of the package ? Is it ok then, to conflict with
    the previous version to not let the user install two versions. I ask
    this because I am really not sure this is supported currently, and
    frankly I don't  really see the interest, and I don't have spare
    time to work on supporting that.

Thank you for your time and advice,

Kind regards,

*Dr. Flavien BRIDAULT*
Director of Software Development
IRCAD France & IRCAD Africa

flavien.bridault at ircad.fr <mailto:flavien.bridault-louchez at ircad.fr>
Tél. : +33 (0)3 88 119 201
		IRCAD France
http://www.ircad.fr/
http://www.ircad.africa/ <http://www.ircad.fr/>

Suivez l'IRCAD sur Facebook 
<http://www.facebook.com/pages/IRCAD/193785273990141>

*IRCAD France*
Hôpitaux Universitaires - 1, place de l'Hôpital - 67091 Strasbourg Cedex 
- FRANCE

Le 17/06/2022 à 21:25, Étienne Mollier a écrit :
> Hi Flavien,
>
> Flavien Bridault, on 2022-06-14:
>> A little "up" in this thread, if someone could have a look at the update of
>> Sight 21.1 and upload it if it looks ok...
> Thanks for the nudge!  Sorry for the delay.
>
>> Le 11/05/2022 à 09:01, Flavien Bridault a écrit :
>>> I updated Sight to 21.1.1 and updated all patches as usual. It looks
>>> good to me for upload.
>>>
>>> Please verify and upload it if you think it is good enough.
> I pulled your changes and ran a build, plus a few tools to help
> review.  Running `lintian --display-info --pedantic`, please
> have a look at the lintian warnings, and see whether they can be
> addressed.  I notably noticed the following:
>
> 	W: libsight: duplicate-font-file usr/share/sight/viz_scene3d/Media/fonts/DejaVuSans.ttf also in (fonts-dejavu-core)
> 	W: libsight: unusual-interpreter python [usr/share/sight/viz_scene3d/Media/materials/genDefaultMaterial.py]
>
> Note that there are many more messages.  You may override items
> if they are not actionable, and document why they are overridden
> in the corresponding lintian-overrides file.  (Note that I
> understood the "elf-error In program headers:" is a lintian bug
> on debug symbols packages, so you may just ignore that for the
> time being in case it crops up with your lintian version.)
>
> You can also run blhc to check the build log, to make sure
> standard Debian build flags are propagated appropriately.  At
> the moment, it reports that flags propagated through LDFLAGS are
> missing in several instances (but maybe there is a good reason
> for that, in which case you can ignore-line-regexp and document
> why).
>
> If you have the opportunity to do so, you may want to address
> the bug #1013036[1]; this is not urgent yet, but will be needed
> before debian 12 release.  After looking at the build log[2] and
> the help page to porting to gcc-12[3], the issue didn't look
> exactly obvious to me though.
>
> [1]:https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1013036
> [2]:http://qa-logs.debian.net/2022/06/09/gcc12/sight_21.0.0-4_unstable_gcc12.log
> [3]:https://gcc.gnu.org/gcc-12/porting_to.html
>
> In general, you may want to have a look at the package tracker
> to have a summary of some of the different issues[4] in the
> existing unstable version you might want to address.
>
> [4]:https://tracker.debian.org/pkg/sight
>
> The package looks otherwise in good shape to me, and I would be
> happy to upload (or happy to be beaten at uploading by someone
> else) once you address the warnings I pointed to and ready the
> debian/changelog for upload to unstable instead of UNRELEASED.
>
> Thank you for your contributions!
>
> Have a nice day,  :)
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://alioth-lists.debian.net/pipermail/debian-med-packaging/attachments/20220628/477bcfcf/attachment-0001.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: XMfcecsg0Gdvs0wW.png
Type: image/png
Size: 6090 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/debian-med-packaging/attachments/20220628/477bcfcf/attachment-0002.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: tfADsBIh1QS7NvDq.png
Type: image/png
Size: 3486 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/debian-med-packaging/attachments/20220628/477bcfcf/attachment-0003.png>