RFS: ed2k-hash

[Tiago Bortoletto Vaz]

On Thu, Sep 10, 2009 at 11:27:27PM +0200, Michael Prokop wrote:
> > Using dh now.
> 
> % grep Build-Depends debian/control
> Build-Depends: debhelper (>= 7), cdbs, autotools-dev, automake
>                                  ^^^^

fixed, thanks.
 
> > > If you fix those issues we could integrate ed2k-hash into the
> > > debian-forensic project and I'd sponsor the package for you.
> 
> > Thanks! Please have a look on it (same address above) and feel free to import
> > to debian-forensic repository. You can change the needed fields in control or
> > anything you may want. I ask you just to let myself as uploader.
> 
> Further investigation in debian/copyright:
> 
> | This is ed2k-hash, packaged by Tiago Bortoletto Vaz <tiago at debian-ba.org> on
> | Sex Ago 28 23:14:54 BRT 2009
>   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^
> Please don't use local settings but instead use a timestamp that
> non-brazilian(?) users understand as well. ;)

sure, fixed.

> | The source used for this package can be found at:
> |        http://edvac.debianbrasil.org/~tiagovaz/
> 
> Please provide upstream's URL, same for debian/watch.
> (Your intention to get rid of debian/ from upstream is fine, though
> that's not relevant for copyright information.)

Agreed and fixed for copyright information. Don't you think is it a problem
having a watch entry pointing to another source but that used for packaging?

> [...]
> 
> | Other's GPL/LGPL sources used in ed2k-hash
> | ==========================================
> |
> |  * Copyright holder: Free Software Foundation
> |  * Files:
> |            - acinclude.m4
> |            - aclocal.m4
> |            - admin/ltcf-gcj.sh
> |            - admin/ltcf-c.sh
> |            - admin/ylwrap
> |            - admin/ltcf-cxx.sh
> |            - admin/missing
> |            - admin/depcomp
> |            - admin/acinclude.m4.in
> |            - admin/ltconfig
> |            - admin/ltconfig
> |            - configure
> |            - libtool
> 
> IMHO that's an unusual format. ;) In the debian-forensics team we
> usually use the Machine-readable debian/copyright, see
> http://dep.debian.net/deps/dep5/

Ok, I'm working on it. Good practice, though.

> But much more import: what about for example admin/am_edit and
> admin/debianrules? Are you sure all the KDE stuff is needed?

The truth is there are a lot of stuff in this upstream tarball which aren't
relevant for the package. The upstream agreed to remove the ./debian, but at
the same time he stated he's not motivated to maintain this tool. I would love
to do a cleanup in this tarball but couldn't find a good solution yet. I
appreciate your opinion here, once I've often faced such situation in Debian.
Options I can imagine:

1) Fork and maintain a new orig source in a better shape;
2) Use a patch system for cleaning up the orig source (AFAIK it can't solve the
./debian/* issue);
3) Bother and bother the upstream to cleanup the source him/herself (and wait
before doing the first upload...);
4) Build a native package.

For ed2k-hash, which we have a very simple tool which will probably not have
updates from the original author anymore, I'd say the first option is the more
suitable. Do you agree?

> In debian/README.Debian you write:
> 
> | Tim-Philipp Müller has agreed to deliver ed2k_hash without the ./debian soon.
> | So, while he doesn't release this new tarball in sourceforge I'll use my own to
> | build ed2k_hash package. The only difference between them is that mine has no
> | ./debian directory inside.
> 
> Though this doesn't look true to me:
> 
> % tar ztf ed2k-hash_0.3.3.orig.tar.gz | grep debian
> ed2k_hash-0.3.3/admin/debianrules
> ed2k_hash-0.3.3/debian/
> ed2k_hash-0.3.3/debian/CVS/
> ed2k_hash-0.3.3/debian/CVS/Root
> ed2k_hash-0.3.3/debian/CVS/Repository
> ed2k_hash-0.3.3/debian/CVS/Entries
> ed2k_hash-0.3.3/debian/changelog
> ed2k_hash-0.3.3/debian/control
> ed2k_hash-0.3.3/debian/rules

Pushed the wrong orig, sorry.

> And IMHO the CVS stuff should disappear from the tarball as well.

Agreed, please help me to decide how to do it (see above), then I make
available a (hopefully) fixed build.

> Thanks for your work.

Thank you,

-- 
Tiago Bortoletto Vaz
http://tiagovaz.org
0xA504FECA - http://pgp.mit.edu