[PATCH] (libclang) Incomplete cursor traversals

[David Blaikie]

On Wed, Mar 25, 2015 at 12:03 AM, jeaye <contact at jeaye.com> wrote:

> On Tue, Mar 24, 2015 at 11:54:17PM -0700, David Blaikie wrote:
> >    [1]
> http://llvm.org/docs/DeveloperPolicy.html#making-and-submitting-a-patch
> >    discussing the rough way to approach this stuff.
> >
> >    Basically -  a patch file attached to an email to cfe-commits with
> "patch"
> >    in the subject would be great. Patches should include changes to the
> test
> >    tree to verify the changes.
> >
> >    - David
>
> So, bugs are ignored and patches can't references bugs?


Patches can/should reference bugs for context, but the easiest way (&
easier => more likely to be dealt with) for a patch to be assessed by a
community member is for it to be attached to an email, sent to a -commits
mailing list, with a test case included in the patch.


> Furthermore, nobody is interested in picking up these changes?


That's quite possible, and certainly a major limitation with LLVM/Clang (&
I assume to some extent, any open source project, though I haven't worked
on many) - if it's not an area of the project that anyone particularly
cares about/is invested in, it tends to rot, patches or not. Those who
haven't worked in the area don't feel compelled to delve into it to
understand the merits or possible issues with the patch.


> Since I'm not interested in delving back into the libclang source, what's
> the best way to just make sure my bugs (which actually contain fixes and
> test applications) don't get ignored?
>

I don't know of a way. Nature of the project is that most people who work
on it have their own goals & work on those - we all do some amount of
community stewardship (reviewing other people's patches, etc) but usually
in a more give-and-take (review other people's patches so ours will be
reviewed) which leaves newcomers or people not interested in general
community involvement, out in the cold.


> There is clearly a set of issues with libclang which have existed for some
> time without any changes; could it be that people submitting bug reports
> (who were kind enough to include patches suggesting how the bugs could be
> fixed) didn't want to jump through hoops to get noticed?
>

Quite possibly, though we do make some effort to encourage those who add
patches to bugs to instead submit them via the commits lists. (not a
perfect process in any case - neither the perfect way to do things nor
necessarily are we perfect about encouraging the use of that imperfect
process)

- David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/cfe-commits/attachments/20150325/45ffe0fa/attachment.html>