[Rpm-metadata] Rpm-metadata Digest, Vol 75, Issue 2

[seth vidal]

On Tue, 2010-08-03 at 10:30 -0600, Matthew Dawkins wrote:
> 

Matt,
 please stop including the full digest in your posts. It's not helpful.

> 
> No time like the present if you ask me, and why force adoption?
> Shouldn't this be all optional? If you want to use lzma give the
> option of --lzma, same for xz or bz2 and gz. Personally I'd like to
> see xz as that's what we use for all our compression. I just said lzma
> b/c I saw it kinda being worked on.


Options just mean a lot more code and units to test and a lot more
places for edge/corner cases to develop.


More code is not good. It's more pain to maintain.

> The same with forcing the adoption of the sqlitedb, why? Just because
> fedora/yum use it??? That's really weak logic. Again, I would say it's
> all about options. Let the repo provider decide.

'forcing'? This is a format of metadata. It's no more about forcing than
any other format forces their choices on anything else.

xml has the advantage of being somewhat extensible. It has the
disadvantage of being an arse-pain to parse and not all that much fun to
generate.

sqlite has the advantage of being available everywhere and significantly
easier to search/interface with.


-sv