On Tue, Mar 31, 2009 at 11:19 AM, Schechter, Michael
> Two things I want to leave you with:
> 1) The behavior issue that I am seeing appears to be related to the atomic commit behavior that
> we discussed last August, and that you fixed in bug 783. It still seems very odd to me that the
> behavior is inconsistent.
They are not really related at all, other than that the directory
layout is the same. The bottom line is that 99%+ of our code uses
Eclipse API for stuff and Eclipse does not know about things outside
the workspace. I think it is reasonable when we are using Eclipse to
crawl your project to stop at the project. Our only alternative would
be to switch to the Java IO classes for this.
The commit change was just a hack where it checks to see if the common
root folder of a commit is a WC or not. That mostly happens in the
SVN code. Hopefully this will become native to SVN in 1.7, as I am
sure the hack is going to stop working then one way or another.
> 2) In the cases where Subclipse correctly detects the bugtraq properties, there is a minor
> problem with the URL it builds. Our bug format is "Issue Number %BUGID%:\n", and when
>Subclipse displays the hyperlink in the history view, it includes the ":" as part of the link
>(TortoiseSVN does not do this). The resulting link tells me that bug "355:" does not exist. Is this an
> issue with Subclipse?
It sounds like it. Can you file an issue with all of your bugtraq
properties, a sample log message and the URL it uses?
To unsubscribe from this discussion, e-mail: [users-unsubscribe_at_subclipse.tigris.org].
Received on 2009-03-31 20:01:37 CEST