> You guys were very quiet about this patch....
Yeah... that wasn't exactly the plan, but there are several not-very-good
reasons for it. The biggest is that, at the moment, the imon-for-Linux
patch is almost certainly unusable, and fam isn't very interesting without
imon. Hopefully in the next day or two I will be able to spend some time
trying it out, but I've been thinking that for about a week now.
> Do the GNOME & KDE folks know about this?
Sort of; I sent mail to a few GNOME & KDE people, and Rebecca Underwood
talked to them at LinuxWorld Expo last month, so they know it's out there,
but if they don't jump on it, it's our fault, not theirs. I'm not
subscribed to any of their development lists, and I'm not actively pushing
it the way it should be pushed. There are all kinds of things we should
be doing, but we can only do them a bit at a time, so it's all going
pretty slowly.
(Hey, the KDE people had support for fam in their KDirWatch class even
before fam was available (!), but I haven't been able to fiddle with it
yet.)
The two biggest things that are missing right now, I think, are a working
imon for Linux, and examples/tutorials on using fam in Qt and GTK+ code.
> If they are interested (and bear in mind that they're resistant to
> linux-only stuff) then I believe this should be pushed along a bit.
Actually, I hope fam won't be linux-only; I talked to a guy last week who
pointed me at an imon-like facility which had been implemented in FreeBSD
by a friend of his, and which is apparently already in the kernel. If fam
is able to make use of that, it might be running with kernel support on
FreeBSD before Linux!
> I'd be interested in working on this in 2-4 weeks time. I suggest that
> the kernel patch should be made really, really thin - much more so than
> at present. Then most of the functionality can be moved into a module.
Any improvements you want to make to the imon patch would be great!
Right now most of what imon does is already in a module, but the way it
hooks into the filesystem code is "sub-optimal" (and there are quite a few
places where it intrudes into the filesystem code). I agree that the less
that slops out of the module, the better. It seems like everything (or
almost everything) imon does should be able to be contained, although I'm
not sure about the start- and stop-exec events, as they're handled
differently.
> Hopefully the Gnome/KDE world has enough manpower to look after the user
> space library.
(which, except for porting to other platforms, shouldn't take much work.)
> If the mainline patch is tiny and buyin from the GUI teams is shown, the
> hard man@transmeta _may_ accept the enabling hooks in the mainstream
> release.
Yes, that's exactly the way I'm hoping it will go.
--Rusty
--
Source code, list archive, and docs: http://oss.sgi.com/projects/fam/
To unsubscribe: echo unsubscribe fam | mail majordomo@xxxxxxxxxxx
|