Re: [gentoo-dev] metadata.xml GLEP for review

[Michał Górny]

On Wed, 16 Mar 2016 20:37:30 +0100
Dirkjan Ochtman <djc@g.o> wrote:

> On Wed, Mar 16, 2016 at 7:43 PM, Michał Górny <mgorny@g.o> wrote:
> > Therefore, I've been slowly writing a proper GLEP that would describe
> > all of metadata.xml in detail. Here's the current draft for review:
>
> Sounds like a good idea!
>
> > https://wiki.gentoo.org/wiki/User:MGorny/GLEP:68
>
> I reviewed your spec based on my experience from trying to create a
> RELAX NG schema for all metadata.xml files that were in the tree at
> the time. I assume you've also validated your spec against what's
> actually being used?

I've validated my XML Schema which conforms to the spec (but isn't 100%
strict on everything). You can see the numeric results in 'backwards
compat' section, also attaching failures.txt with specific list in
case you wanted to look through it.

> I have a few questions:
>
> - I had the upstream maintainer's email element pegged as mandatory.
> Don't you think that makes sense? A name-only maintainer element seems
> relatively low-value to me.

As for upstream stuff, I tried to keep it as GLEP 46 says. In this
case, it's unclear whether it's supposed to be mandatory. There's no
explicit notion either way. However, as I pointed out, there are 4
packages which list maintainers with empty e-mail address. If we are to
consider email obligatory, we should drop maintainer entries from them.

> - You list a number of the upstream child elements (changelog, doc,
> bug-to) as "zero or more". Doesn't it make sense to make (some of)
> these zero or one?

This is another problem with GLEP 46 -- I have no clue. After
discussing this with ulm, he suggested we use the same number for all
elements to reduce possible confusion.

Maybe it would be useful to query GLEP 46 authors but sadly I don't
have time to do that right now.

-- 
Best regards,
Michał Górny
<http://dev.gentoo.org/~mgorny/>