[Sugar-devel] Adding information about the repository to activity.info file
James Cameron
quozl at laptop.org
Fri Dec 26 15:24:33 EST 2014
On Fri, Dec 26, 2014 at 09:45:33AM -0300, Gonzalo Odiard wrote:
>
> On Tue, Dec 23, 2014 at 5:36 PM, James Cameron <[1]quozl at laptop.org> wrote:
>
> On Tue, Dec 23, 2014 at 03:01:10PM -0500, Walter Bender wrote:
> > Can I suggest that if we are going to touch the .info files, we also
> > consider adding the maximum participants field (it is probably
> > possible to compute from a simple examination of the code -- maybe a
> > good GCI task). Having this in [2]activity.info makes it easier to test
> > for in the Sugar Shell, allowing us to finally honor the limit on
> > participants suggested by activity.max_participants.
> >
> > Any other changes to [3]activity.info we should be considering now?
>
> Just to get the conversation started ...
>
> - where to report bugs.
>
> - where to look for a new release.
>
> Is not the repository enough?
It often isn't clear, by looking at a repository, when a release
happened. So some metadata is needed somewhere; either a convention
of tags, commit message, or tar.gz for packaging.
> - what processor architectures are supported by the binaries enclosed.
>
> This field would be optional, only needed for activities with binaries.
> What wold be the options? (i386, x86_64, arm??)
I don't know a standard way to describe those options.
> - list of versions of Sugar certified by the activity.
>
> Maybe sugar_version_min and sugar_version_max both optional?
I don't think a range is flexible enough. A list is more flexible.
By certified I mean a developer has fully tested the activity on the
version of Sugar.
Not intended to block the activity though, but rather to inform.
> - list of distributions certified by the activity.
>
> This is difficult to know for the activity developer.
Yes, but it would have at least one entry.
> - what distribution packages (beyond the Sugar standard list) are
> required by the activity.
>
> Again, haw can we solve the problem of dependencies packaged with different
> names
> in different distros? Is this a problem today or package names are mostly
> standard?
Correct, there is no distribution agnostic standard. So package names
would have to be qualified by distribution name.
--
James Cameron
http://quozl.linux.org.au/
More information about the Sugar-devel
mailing list