On Wed, 2005-07-13 at 11:14 -0400, Daniel Veillard wrote: > Following the inclusion of inotify in Linus kernel and the change of > kernel API, I made a new release freezing the current state of CVS, > the changes from 0.1.1 are: > - inotify patches from John McCutchan > - a number of the server code cleanup patches from Neal H. Walfield > (but not all of them) > > Depending on people's interest we may do different things: > 1/ try to stabilize the 0.1.x branch based on the upcoming feedback > 2/ use this transition at the lower level as an opportunity to finish > revamping the server code (i.e. go with the other patches from > Neal) > > I would be fine going through plan 2/ assuming Neal is interested in > continuing the cleanup. We could also branch, but I would find that a bit > too heavyweight considering were not even yet at an 1.0 release. >
If it's worth anything, I say go for plan 2. > What do people think ? I'm still pondering the issue of watching all > ancestors from a monitored node, I also need to start playing with inotify, > it certainly have different behaviour from dnotify, and probably can > help detecting change on any of the parent without the same crazyness as > dnotify forces us to. > At this point dnotify is legacy, and I'm fine optimizing for the inotify > case as long as we don't break the simpler dnotify support and other kernel > mechanism on BSD/Mach/etc... Excellent! I'm happy you are being so supportive of inotify now that it is included in the kernel. I agree that we should continue to keep the kernel helper backends as modular as possible. Let me know if you have any questions about inotify. _______________________________________________ Gamin-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/gamin-list
