Re: Request for write access on OctaveForge server

[Mike Miller]

On Sat, Oct 31, 2015 at 11:09:46 +0100, Julien Bect wrote:
> Perhaps the first step towards reducing the release delay (and reducing
> Carnë's workload) would be to fully document those steps (which probably
> include some basic sanity checks, pushing the tarballs to the release file
> system, pushing the documentation to the web site, etc.).
> 
> Then we would have two options : either expanding the "release team" (which
> is actually composed of Carnë only) as you suggest, or automating those
> steps (in which case we could have some sort of "release deamon" that does
> what Carnë actually does, but I have no idea if this is realistic).

I agree with automating as much of the release process as possible. Some
concrete steps that any of us could take to help with this would be
writing scripts to do QA checks on package releases. Oliver's maintainer
makefile approach to making package releases has helped a lot in that
direction, but other independent checks are needed to catch oversights
and common mistakes in packages and associated docs.

Some examples off the top of my head:

 * .hgignore files
 * binary oct-files or others
 * build failures
 * test suite failures
 * too tight or loose file permissions
 * missing, empty, or black plots in docs
 * DESCRIPTION, INDEX, or NEWS files not updated
 * mistakes in dependencies
 * doesn't work with older or newer Octave versions or other packages

-- 
mike