[postgis-devel] repository hosting

[Greg Troxel]

Sandro Santilli <strk at keybit.net> writes:

> Actually, I don't see it as a huge benefit myself.
> Maybe packagers do, if I'm not wrong Greg Toxel mentioned it would
> be easier for them to track PostGIS from the debian git repo.

I am pretty sure I didn't say that.  My strategy has been to file
postgis bugs for anything that doesn't just build without patches, and
to carry patch files (literally a file with a patch that is applied
after unpacking and before building).    Sometimes I look at other
packaging systems, but in a healthy ecosystem those people file bugs
upstream and upstream fixes them, so I get them anyway.

pkgsrc is stored in CVS, but we don't keep a copy of postgis -- just a
few control files that say how to build it and what the list of
installed files is.

What I really care about as a packager is

  having actual release tarballs
  having no churn in extensions, how it unpacks, etc.
  having portable code
  fixing portability bugs quickly so I don't have to carry patches
  having a NEWS file to summarize package updates

Secondarily, when packaging things, I tend to end up contributing to the
package itself to resolve some of the portability issues or other bugs.
Here, my opinions are really just as another contributor.

  avoiding agreements with companies, especially those involving indemnification
  avoiding cookies from advertising companies
  I prefer git, because I can carry local commits and have local history.


I see a move to git as not primarily benefitting those with commit
privileges.  Those people can commit changes when they judge them to be
sound, so they don't tend to carry changes.  The benefit of git is in
keeping changes from unauthorized committers (almost all people, and the
pool of future committers) within the tool.  When I work on NetBSD, I
will often get a patch from people.  99% it does not come with a
fully-written suitable commit message.  With git, the unit of what you
review is a commit, and the message is part of that.  So the real point
is that git separates authority to make changes and ability to prepare a
complete change.

The other point is what tools new people are familiar with.   That
really seems to be git.

All that said, I have not found the use of svn or submitting patches in
tickets to be a problem for me.  People (strk mostly for my patches)
have responded quickly and committed them.   So while I think
osgeo-hosted git is better, I don't think it matters that much.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 180 bytes
Desc: not available
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20151016/62ceb07c/attachment.sig>