Open Stack

On 09/01/2014 01:38 AM, Ian Wienand wrote:
> On 08/29/2014 10:42 PM, Sean Dague wrote:
>> I'm actually kind of convinced now that none of these approaches are
>> what we need, and that we should instead have a .bashateignore file in
>> the root dir for the project instead, which would be regex that would
>> match files or directories to throw out of the walk.
> 
> Dean's idea of reading .gitignore might be good.
> 
> I had a quick poke at git dir.c:match_pathspec_item() and sort of came
> up with something similar [2] which roughly follows that and then only
> matches on files that have a shell-script mimetype; which I feel is
> probably sane for a default implementation.
> 
> IMO devstack should just generate it's own file-list to pass in for
> checking and bashate shouldn't have special guessing code for it
> 
> It all feels a bit like a solution looking for a problem.  Making
> bashate only work on a passed-in list of files and leaving generating
> those files up to the test infrastructure is probably would probably
> best the best KISS choice...
> 
> -i
> 
> [1] https://github.com/git/git/blob/master/dir.c#L216
> [2] https://review.openstack.org/#/c/117425/

Sure, I think part of this is we built the tool inside of devstack, and
now in the base case it's really awkward to use with devstack. I'd say
that's part of what I feel is the 0.x lifecycle for bashate, figure out
how to live as a separate tool and not add a ton of work to our existing
projects.

One of the things that could make it better is to add file extensions to
all shell files in devstack. This would also solve the issue of gerrit
not syntax highlighting most of the files. If people are up for that,
I'll propose a rename patch to get us there. Then dumping the special
bashate discover bits is simple.

	-Sean

-- 
Sean Dague
http://dague.net