[webkit-dev] [Fwd: Fwd: Re: Moving forward with WebKit/GTK+]

[Maciej Stachowiak]

On Nov 3, 2008, at 3:15 AM, Pierre-Luc Beaudoin wrote:

>
> I'm only an outsider to the WebKit project, but I wonder if it would
> make sense to have less strict rules, or a branch with less strict  
> rules
> for WebKit/GTK+ -- since WebKit/GTK+ is a bit young and needs some  
> work,
> it might make sense to be more permissive so that development can  
> occur
> at a faster pace. I could totally be wrong, though :-)
> (maybe having more reviewers, as already suggested, is enough)

Here are some things I would recommend to make WebKit/Gtk+ move faster:

1) Separate API review from code review. There are many patches that  
could be reviewed at a technical level by reviewers who work on other  
ports, but since they contain API changes, others are reluctant to say  
anything. I would suggest setting up a mailing list, post proposed API  
changes there, and let them be official after a week if there was no  
objection. If bugs are tagged as being API reviewed (by whatever  
process), then others could focus on technical and style issues during  
review. I think having some sort of lightweight review process is  
probably better than making API a free-for-all.

2) Finish up the Gtk port of DumpRenderTree, and start running layout  
tests on the WebKit/Gtk+ buildbot. It's great that there is a buildbot  
for the Gtk port - it allows others to keep the build working. I think  
it would be even better if the regression tests could run, because (a)  
it would be a bit less scary to review Gtk patches, since we'd at  
least know the regression tests still pass; and (b) others could help  
ensure that core changes don't unexpectedly cause functional  
regressions in the Gtk port.

Regards,
Maciej