Bug#338789: cogito packages
On Tue, 2005-11-15 at 17:36 +0000, Gerrit Pape wrote:
> Hi Andres,
>
> On Mon, Nov 14, 2005 at 01:56:30PM -0500, Andres Salomon wrote:
> > I'm also interested in taking it. Perhaps we could comaintain it and
> > git-core? It's quite important for the kernel team (and others) to have
> > this package.
>
> sure, let's see first whether we agree upon how the package(s) should
> look like. Here's what I plan to do:
>
> o split the package into git-core, cogito, and also split off gitk:
> source git-core with binary packages git-core, gitk; source cogito
> with binary package cogito
> o maybe split off git-cvs*, git-svn* also
> o don't link against openssl: set NO_OPENSSL, disable --merge-order
> in rev-list.c; link against libcurl3-gnutls (#321301)
> o fix depends and change some depends into recommends
Sounds good so far...
> o complete repackaging without debhelper or any other build system, I
> usually use a simple collection of implicit Makefile rules for my
> packages
This I disagree w/; debhelper is quite valuable, and keeps a lot of the
manual labor involved w/ package maintenance to a minimum.
> o try to solve the git<->git conflict with gnu interactive tools, for
> now keep the name 'gt' for the git wrapper
I was thinking of just doing a conflicts w/ git; it would violate
policy, but it should only be temporary. I wouldn't want to see
git-core with 'gt' enter etch, have people get used to it, and then have
to wait for git-core with 'git' to get into etch. Better to just keep
the same name, and leave it out of etch for the time being (imo).
>
> Packaging generally shouldn't be that hard, I'll prepare packages and
> post a link here, so you can take a look at it, ok?
>
> Regards, Gerrit.
Reply to: