Re: RFS: mscgen
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Y Giridhar Appaji Nag wrote:
> Hi Niels,
>
> I looked at mscgen 0.16~svn35-1 on mentors.debian.net
>
> On 09/07/02 16:51 +0530, Y Giridhar Appaji Nag said ...
>> [...]
>
> Comments:
>
> I noticed that the "released" version upstream is 0.15 and you choose to
> package the svn version. Is this a conscious decision? Any reason why you
> choose to do it this way? The svn snapshot is suitable for widespread user.
> You might want to upload 0.15 to unstable and upload 0.16~svn to experimental.
>
> For an svn snapshot package, the debian/watch file doesn't really make sense.
>
Yes, the 0.15 release suffers from two very apparent bugs (an easily
introduced seg. fault and a regression on a "promoted" sample file). I
think that these could, however, easily be changed into using being a
0.15 with some debian patches.
Though such a change would require that the version number of mscgen was
lowered. How would I go about doing that in regards to my changelog?
Merge all entries so far into 0.15-5 (updating svn-releases to patches)?
I think it would also greatly simplify the the watch-file if I had used
0.15+svn[NN] rather than 0.16~svn[NN].
> in debian/control, it doesn't look like you need versioned dependencies on
> most of the packages in Build-Depends. You don't need to Build-Depend on
> libc6 (Hint: see build-essential) [...]
Okay, fixed these.
> and you probably need libgd2-noxpm-dev or libgd2-xpm-dev but not both.
>
As far as I can tell that is what it says in my control file, though it
was not nice of me to writing like: "depA | <newline> depB".
Nevertheless, I am considering to use the virtual package "libgd-dev"
instead.
> in the Package: mscgen stanza, you don't need "libgd2-xpm | libgd2-noxpm" in
> Depends. That would be generated automagically from ${shlibs:Depends}
> depending on what you have in Build-Depends based on the above (Hint: read
> about substvars and ${shlibs:Depends}).
>
> In the Description "Can be used ..." is a bit abrupt, you might want to say
> "mscgen can be used ..." and "There also exists extensions ..." would rather
> be "extensions also exist ...". "although MSCs need not be complicated to
> create or use." is a bit odd in there.
>
> The description of the dpatch 01_debian-patch is verbose and is incorrect wrt
> what the patch actually does (in particular about the linker flags).
>
> in debian/rules, it is not necessary to have "dh_installdocs TODO" when there
> is a debian/docs file that lists TODO. dh_makeshlibs is not needed.
>
Fixed these.
> Suggestions:
>
> In debian/control Priority need not be extra, you can make it optional. See
> the thread http://lists.debian.org/debian-mentors/2009/05/msg00666.html also
>
Done.
> Please consider maintaining the package in collab-maint on alioth, you would
> want to add the Vcs-* headers to debian/control in case you do that.
>
> In case there is no location from where you can download the upstream source
> (because you are packaging an svn snapshot), it is a good idea to have a
> get-orig-source target in debian/rules that gets the upstream source from svn
> and makes an .orig.tar.gz.
>
Both of these sounds like good ideas, I will have a look at them.
> Cheers,
>
> Giridhar
>
I will wait with uploading the corrections until the "version" number of
this package as been sorted out.
~Niels
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
iEYEARECAAYFAkpVvm8ACgkQVCqoiq1YlqwK5gCgtiTpjiL+KqEUWVYTI9nvbMgv
CIMAn2a71fcCTLJI4HtOBlvg+XTTiAyN
=g59C
-----END PGP SIGNATURE-----
Reply to: