Community
Participate
Working Groups
Created attachment 119673 [details] Configuration log file I installed I20081202 on top of eclipse-SDK-I20081125-0840-win32.zip from the Eclipse update site. The install succeeded but bundles in my target were not found causing my workspace to be unusable (Simon says: no spaces in workspace path) (probably a PDE problem). I tried reverting back but now I can't start Eclipse at all. I have a number of plug-ins loaded on top of Eclipse (e.g. EMF, GEF) and some plug-ins were loaded using the links folder. I don't know if it is a factor, but when I reverted, I clicked Finish and was prompted to continue. The wizard stayed up and I hit Finish again and was again prompted.
The moral of the story is don't revert, even if Simon says ;-)
In retrospect it was a bad idea to revert given the recent path changes. I suspect that this set of changes (pathing) will make rolling back from a 3.5 install to 3.4 tough given that we had to correct 3.4 behavior in several cases. A few additional things I noticed in the reverted install: 1) the eclipse.ini had lost the -startup 2) the launcher jar was refereneced using an absolute URI (should be a file path) 3) the bundle.info was written with relative uri instead of the older style file:... bundle locations we should be writing for an older simpleconfigurator.
we need to understand the scope of this while testing m4. For example, can I revert to a previous I build leading to M4?
It seems that some simple revert is possible. Moving to M5.
For the problem reverting the bundle info file (3) one thing that is causing problems is that the old version we're matching against is too old. We should bump it to any versions prior to M4 to allow us to for example revert to M3 in the simplest case. This safe since even for the builds leading up to M4 since we can still read old style relative URI in bundle.info.
Update on this bug... Of the 3 problems I listed in comment 2 only the first remains. I'm going to mark this bug as a dup of bug Bug 258879 which is the first problem e.g. "-startup". The other 2 are already fixed in HEAD. *** This bug has been marked as a duplicate of bug 253862 ***
The dup'ed bug is right. Above I meant to reference bug 253862.