On 10/20/07, Torsten Curdt <[EMAIL PROTECTED]> wrote: > Hey Niall, > > Thanks for looking into that. But frankly speaking I am not a big fan > of that for a couple of reasons. > > That would make it the 3rd FAM implementation in the commons > codebase. There already is commons-jci-fam, the commons-vfs fam and > then we would also have the commons-io fam.
VFS could replace IO completely but the power of VFS comes at the cost of a further layer of abstraction and theres still value in having a simpler, "closer to the metal" Commons IO when all thats required is local file system functionality. I think thats both true for VFS and IO as a whole and for fam in particular. So two implementations is no bad thing IMO - but I agree that three implementations isn't desirable and my hope is/was that if what I proposed was accepted that JCI would migrate to the IO impl. > With commons-jci-fam we already have a well tested and released > version as own artifact. So I am not really seeing the need for > making the general commons-io bigger while this is a more specific > need that IMO maybe justifies it's own artifact. I would be more than > happy to accept patches and improvements within the JCI codebase - > that's for sure! I also don't think that its big enough for its own component - the core of it is only 2 interfaces and 4 classes - that doesn't seem too big for IO to me and I think its currently "hidden" in JCI where people wouldn't expect to go looking for something like this - whereas I think they would in IO. Your point about patches/improvements for jci-fam is well taken though :) The only problem I see is the scope for breaking/changing the already released API (it was far easier not having to worry about the existing code/API by putting forward something new to IO). > Mario and I talked about merging the FAM implementations (although I > fear that's not easy). We even considered a commons-fam > project ...it's just that the codebase is a little small on its own. > > Maybe it would make more sense to do these changes in commons-jci-fam > first and then move that to a new commons-fam? Sounds reasonable, although I still think IO is a better home for fam than a separate component. I'll try and find some cycles for this, but not sure when. The main difference is in the FileObserverImpl so if you have the time/inclination then feedback on whether you think it has merit and worth factoring back into FilesystemAlterationObserverImpl would be appreciated. If not, no worries and I'll work up a patch at some point. Niall > WDYT? Niall? Mario? > > cheers > -- > Torsten > > > On 20.10.2007, at 08:50, Niall Pemberton wrote: > > > I have created an issue ticket to add the functionality (with some > > refactoring) provided by the fam module in JCI to Commons IO: > > > > https://issues.apache.org/jira/browse/IO-132 > > > > If its accepted by Commons IO I hope it can be used by JCI as well - > > rather than duplicating. > > > > Comment welcome > > > > Niall > > --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
