This Bugzilla instance is a read-only archive of historic NetBeans bug reports. To report a bug in NetBeans please follow the project's instructions for reporting issues.
I have found a problem in filesystem implementation. It scales from MultiFS to JavaFS, so it is up to you to find out what is the actual problem. To reproduce my observation: 1. mount nb_all using JavaCVS (all modules) 2. mount a treefs at openide/src 3. be sure that all sources in openide/src are compiled Select org/openide and invoke Clean. The action will take for ages
Created attachment 4979 [details] Few stack traces...
added PERFORMANCE keyword
You are right, it`s almost unusable. The problem is obvious. First JavaCvsFileSystem$ListImpl.children is really slow. Second and probably more serious is that such slow children are invoked from MFS.updateAll again and again (after delete - from its parent whole hieararchy down is refreshed). Delete of every file takes few seconds. Perhaps it`s not necessary to refresh whole hierarchy after delete deep down in case that FileEvent.getFile ().isData () (the same for fileDataCreated). Also refreshFolder where map == null (not known children yet) should not bring in life new FileObjects if not necessary, then the number of refreshed FileObjects should not grow as rapidly as it grows at the moment.
Fixed in trunk (MultiFileObject.java 1.94). There is not necessary to update whole hierarchy down if deleted FileObject is not folder. Now is much better. (Also in updateAll is not necessary to refresh delegates - will be done with #20528)
verified - dev3.4 build #20020325
Resolved for 3.4.x or earlier, no new info since then -> closing.