[Ubuntu] creating GRASS packages for UbuntuGIS

[Bas Couwenberg]

On 2016-01-14 12:17, Martin Landa wrote:
> 2016-01-13 23:04 GMT+01:00 Bas Couwenberg <sebastic at xs4all.nl>:
>> When the packaging lives on a non-default branch (i.e. not master), 
>> you need
>> to document the branch in the Vcs-Git URL and gbp.conf. See this 
>> commit for
>> example:
>> 
>> 
>> https://anonscm.debian.org/cgit/pkg-grass/grass.git/commit/?h=experimental&id=a32ba58d0e41e41b3c4608a1ee4d926fecc96d18
> 
> OK, done for ubuntugis/wily in [1]. I was just surprised that `git
> commit` also push changes to the origin. So `git push` is not needed,
> right?

No, `git commit` only commits in your local repository, to share the 
changes with the repository on Alioth used by others you need to push 
your local changes with `git push`, this is one of big distinctions 
between Subversion and Git. With SVN you commit to the shared repo, with 
git you commit locally. This allows you to rebase your local changes 
into coherent commits before pushing them to the shared repository to 
ease merging and review in the shared repo.

You should always push your changes before or directly after uploading 
the package (and tagging the release).

https://pkg-grass.alioth.debian.org/policy/packaging.html#git-push-to-alioth

>> Did you notice the lintian warnings in your build log?
> 
> No, I checked for lintian and didn't find anything about warning? What
> do you mean exactly?

Review the lintian tags in your log:

  cat log.packager | awk '/+++ lintian output +++/,/+++ end of lintian 
output +++/' | grep -v ^+++ | less

Optionally filter out the overrides and their comments by adding ' | 
egrep -v "^(N|O):" before the pipe to less.

You'll see this warning for the various binary packages:

  W: grass-core: possible-new-upstream-release-without-new-version

You can lookup the extended description online, which I linked in my 
previous mail.

All lintian issues of severity info (I:) and higher need to be reviewed 
and fixed before the upload (or overridden if it's a false positive, or 
purposely left unaddressed, a comment should always be added to explain 
the override).

Kind Regards,

Bas