Community
Participate
Working Groups
The directory involved is /home/data/cvs/tools/org.eclipse.orbit/org.eclipse.orbit.releng At some point, between 12 21 07 17:01 and 12 22 07 13:04 the branch called 'set1' seems to have been "reset" to be equal to the 'head' branch. At first I assumed some "user error" where some committer accidentally checked in the wrong code to the wrong branch, but looking at the cvs change log and history (as best I know how) I couldn't seen any indication that the files had been changed/deleted, etc ... at least with a cvs client, which would have left a trail. And I don't think any committers would be messing with the file system directly. In Orbit, since we tag this directory every build, I could see the cvs tagged version vI200712211701 was still there, and seemed correct, on the surface. So, to get things working again, I "restored" the branch to that version of files. So ... there's nothing blocking us, or needed immediately from Eclipse Foundation staff. But I'm opening this bug to track the issue, and if possible, perhaps webmasters could check some logs, or something, that might indicate what happened?
I found this entry in the CVS history: cvs -d /cvsroot/tools hi -a -D 2007-12-19 -xT | grep orbit T 2007-12-21 22:25 +0000 skaegi org.eclipse.orbit/org.eclipse.orbit.releng [set1:HEAD] Looks like skaegi is responsible for this.
I'll go ahead and close this as WORKSFORME, as the branch hasn't mysteriously disappeared. I'm guessing the Eclipse CVS client cannot poll for the same history as the command-line, as this type of change is only recorded in the CVSROOT/history, not in each file change.
Thanks for the investigation, Denis. We'll follow up with Mr. skaegi :) I am relieved to learn there is at least an explanation for it. And, looks like I have more to learn about CVS from the command line!
