[PATCH] kdelibs kio::KDirWatch

Josef Weidendorfer Josef.Weidendorfer at gmx.de
Wed Jul 18 19:08:42 BST 2007


On Wednesday 18 July 2007, Flavio Castelli wrote:
> On Wednesday 18 July 2007 16:09:38 you wrote:
> > AFAICT the bug is about not supporting watchFiles and recursive flags in
> > KDirWatch::addDir(). But your changes do not touch this function?
> My patch implements watchFiles option. So KDirWatch::addDir() can get dirty() 
> signal when file system events happens inside a watched dir.

Ah, fine.

> In the next future it will be possible to enable also the recursive option, 
> but these changes have to be applied.
> 
> BTW, before applying my patch I wasn't able to get KDirWatch working. My small 
> test program didn't receive created(), deleted() and dirty() signals... Now 
> everythings works fine.

I suppose with inotify?

> > The benefit of a separate thread would be to prohibit deadlock situations.
> > However, I am not really sure what your changes fix in this regard? Or
> > is this kind of a cleanup to prepare for future stuff?
> It's a cleanup in preparation for future goodies, forgive me if I didn't 
> explain myself before.

Cool.

> > > Talking about kqueue support, I've still some bugs to fix and some small
> > > problems to solve.
> > > I'll commit this new feature as soon as possible. Changes regard only
> > > KDirWatchPrivate and take place inside kdirwatch.cpp . Kqueue support
> > > will be added in a way similar to inotify one. Since neither ABI nor API
> > > will be broken I think the commit can be done also after API freeze. Am I
> > > wrong?
> >
> > Yes.
> Ok, thank you for the warning. I won't do it ;)

Oh, sorry. Misunderstanding. As long as every change is only about KDirWatchPrivate,
it is ABI/API compatible, and is safe even after the freeze.

Josef




More information about the kde-core-devel mailing list