[Bug 478581] Package Review: pnglite - A lightweight C library for loading PNG images

[bugzilla at redhat.com]

Please do not reply directly to this email. All additional
comments should be made in the comments box of this bug.


https://bugzilla.redhat.com/show_bug.cgi?id=478581


Robert Scheck <redhat at linuxnetz.de> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |ASSIGNED
                 CC|                            |redhat-bugzilla at linuxnetz.d
                   |                            |e
         AssignedTo|nobody at fedoraproject.org    |redhat-bugzilla at linuxnetz.d
                   |                            |e




--- Comment #1 from Robert Scheck <redhat at linuxnetz.de>  2009-01-01 13:41:17 EDT ---
[16:37:47] < lkundrak> cassmodiah: I'm wondering if you can review pnglite?
[16:37:57] < lkundrak> cassmodiah: should be fairly simple -- just one .c file
:)
[16:38:46] < cassmodiah> i'm not sure with this
[16:38:58] < cassmodiah> wait, i know which is a better guy for that
[16:39:00] < cassmodiah> rsc ping
[16:39:14] < lkundrak> :)
[16:44:20] < rsc> cassmodiah: pong
[16:46:02] < cassmodiah> rsc do you have time for a short look at a review
request?
[16:47:05] < cassmodiah> https://bugzilla.redhat.com/show_bug.cgi?id=478581
[16:47:06] < buggbot> Bug 478581: medium, low, ---, nobody at fedoraproject.org,
NEW, Package Review: pnglite - A lightweight C library for loading PNG images
[16:49:44] < rsc> lkundrak: I'm not absolutely sure, whether the ABI soname
stuff is clever. Was this practised somewhere else was well?
[16:51:17] < lkundrak> rsc: I'm not really sure. I know I did that, but don't
remember if it was in Fedora package. But if upstream is not willing to
maintain soname (which I did not ask though), is there anything else we can do?
[16:53:16] < lkundrak> rsc: you know -- pnglite is just a single pnglite.c
file, upstream doesn't even maintian a build script where soname could be
enforced
[16:54:52] < rsc> lkundrak: understood. Why is %{abi_minor} == %{release}? If
you've to rebuild for e.g. a new GCC; soname would be minorly bumped...
[16:55:31] < lkundrak> rsc: abi_minor was not meant to be the part of soname.
If it is, then it is a mistake
[16:56:04] < lkundrak> rsc: -Wl,--soname,libpnglite.so.%{abi_major}
[16:56:10] < lkundrak> rsc: there's only abi_major
[16:56:12] < rsc> libpnglite.so.%{abi_major}.%{abi_minor}
[16:56:24] < lkundrak> rsc: right, that's not a soname. only a filename.
[16:56:35] < rsc> ah right. Sorry.
[16:57:43] < rsc> but anyway: If you bump %{version}, %{release} should be
reset. Then you would have to increase %{abi_major} to ensure, that
libpnglite.so.%{abi_major}.%{abi_minor} is newer, not older from the filename.
[16:57:53] < rsc> is that expected behaviour?
[17:00:18] < lkundrak> rsc: yes. that's why there's a huge comment in the
beginning
[17:00:54] < lkundrak> rsc: but in case a new major version is released without
abi change, then we would unnecessarily change soname
[17:01:05] < lkundrak> rsc: I don't think that's likely though
[17:11:03] < lkundrak> rsc: on a second thought -- I don't know what is the
last digit in the library file name for. Probably it won't be a problem if it
was reset. Probably we could decouple if from release.
[17:17:59] < cassmodiah> lkundrak one of my teeworlds testers has no sound
[17:19:26] < lkundrak> cassmodiah: well, I do not think it is related to any
change we did
[17:20:34] < rsc> lkundrak: I'll try the formal review in a few minutes

-- 
Configure bugmail: https://bugzilla.redhat.com/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.