>
> Yes, that is the error I was talking about earlier. I
> suspect it is in code that we can just remove now. Martin is
> going to cleanup the old code that he obsoleted. Although
> you may be right that some of the new stuff still needs
> serialize code added?
>
> Yes, Synch is looking pretty good now.
>
> Mark
Well,
the cleanup patches are here.
What I did:
- removed now obsolete StatusCacheComposite
- removed option to save cache status from code and GUI
- fixed wrong handling of unversioned resources
- fixed the svn:ignore syncView issues
- fixed the right-click delay
- fixed the syncView problem when deleting whole trees of directories
+ implemented getResource() method for LocalResourceStatus so it now
properly determines the resource kind (either from ResourceInfo or
svnstatus.kind)
- removed GetStatusCommand#getResource and replaced it's calls by direct
LocalResourceStatus#getResource()
There are still some issues to be resolved, but most of them are related to
performance, not correctness anymore ...
Just for the record:
- sharing already existing project does not work properly (the sync info is
get before the wizard asks for it and the .svn files are not flagged as
team_private)
- during checkout as and also many other times the "svn status" is invoked
too many times on e.g. sibling resources. Proper use of recursive update
strategy should be considered here.
- there are some error message in SVNWorkspaceSubscriber which are not
translated in the .properites file.
- when directory property change is reverted (e.g. add to svn:ignore), the
child resources (e.g. the ignored file) are not refreshed after revert
- when project is disconnected/unshared to status infos are not removed from
resource#resourceInfo's.
- seems like .svn resources are refreshed too many times, might be a problem
with team_private flags ...
- seems like some listeners are notified unecesarily several times
Besides, the synchronization stuff seems to be relatively stable and correct
now in general.
At least in my workspace and my testing scenarios.
Further testing and feedback from everybody is of course very welcome ...
Martin
Received on Mon Jul 11 00:52:04 2005