https://bugzilla.redhat.com/show_bug.cgi?id=1110386
--- Comment #4 from František Dvořák <valtri(a)civ.zcu.cz> ---
(In reply to Richard Shaw from comment #3)
> Maybe it is not needed to instal it, or alternativelly you can use GPLv2
> license for the -example subpackage? In eihter way, you can ask upstream
> about licensing: they are intended to use LPGL for codec2 project and maybe
> they would rather relicense the menu.sh file under the same licesne?
I'll ask but I may just change it. While I am not specifically a programmer,
I am a contributor and have commit access to upstream as I wrote and now
maintain their cmake build system.
OK. Maybe there can be needed the original author to confirm? But it is only a
small script anyway contribued to LGPL project...
>
> For example (release field): 1.20140616svn1657
The is more or less 0.3 official (as it gets) we don't yet do formal release
archives. The irony of the guidelines is that svn1657 tells you exactly what
svn revision it is, but it's not required, but the date of the checkout is
required. I disagree with the guidelines here but the rules are the rules so
I'll fix it.
OK, thanks.
> 6) Is the build inside build_linux needed? (I guess you use it because it is
> mentioned in READMEs?)
Both cmake and I prefer out of source builds. It's in the readme because I
wrote it. :)
OK, I understand. :-) Plus cmake is not able to do distclean, right?
> 7) It is recommended to track major version of the libraries in %files, like:
>
> %{_libdir}/libcodec2.so.0
> %{_libdir}/libcodec2.so.0.*
I can change it if you feel strongly about it but I actually manage the
soversion so I'm not too worried.
OK. You know about it already, it's up to you what to prefer. :-)
> 8) ChangeLog, NEW, AUTHORS are empty files, they're not needed to install
I usually include them even if they are empty because they may not always be
empty and who actually checks for that during the update process?
Rpmlint cries loudly about it. But if you have it in other packages too, I'm
not for breaking uniformity.
> 9) README.cmake is only about build instrutions and it is not needed
I should have caught that since I wrote it :)
:-)
> 10) codec2-examples should have dependency on "%{name} =
> %{version}-%{release}", not the -devel. Was there any reason for that?
Maybe I should rename the package to codec2-devel-examples? It's not needed
for the main package at all.
It's not needed IMHO. I think users will understand codec2-examples is not
needed for functionality of the codec.
It is only the Require, I think there should be "%{name} =
%{version}-%{release}" instead of "%{name}-devel =
%{version}-%{release}"...
> Enhancements:
>
>
> 14) Is possible to use something in %check?
>
> There is a testsuite, but if I understand corretly, it is not intended for
> automatic testing. (it is more for codec developers?) It could be commented
> in the .spec file that the testsuite exists, but it can't be used.
It's not designed to be automated yet. My plan is to add automatic testing
via ctest but the current developer is more interested in codec2-dev than
codec2 so I'm not sure if it will happen anytime soon.
In fact, I'm moving the binaries to the devel package as they are not
terribly useful except for development (and testing) purposes.
Maybe examples could be better place? codec2-devel subpackage could be for
developers using the codec2, or BR for other packages.
> 15) man-pages: there is recommended (but not required) man-page for each
> binary in /usr/bin
I would like them to have one as well but I doubt I can get upstream to
write them.
That's not so strong excuse since you're upstream co-maintainer. ;-) But it's
OK. :-)
I'll try to get a new SRPM and SPEC posted today if I have time.
--
You are receiving this mail because:
You are on the CC list for the bug.
You are always notified about changes to this product and component