Community
Participate
Working Groups
I20051018-0800, GTK+ 2.6.8, KDE 3.4.1, X.org 6.8.2, Linux 2.6.13 + When I saw Bug 113454, I started up Eclipse and synchronized with HEAD. Even with all the code from HEAD, no errors showed up in my Problems view. + So, I opened the file referenced in the first compile error (ResourceMappingMarkersTest), and it had several errors in it. + I clicked on one of the methods that was an error (problemView.getCurrentMarkers()), and hit "F3". Now one error appears in the Problems view: "ModelProvider cannot be resolved" in CompositeResourceMapping. + I tried cleaning all projects, to see if that would motivate the errors to appear. The one error that was there previously then disappeared. I had a hard time deciding whether to make this "blocker" or "major". The big problem, as I see it, is that this can lead to broken builds (as we've seen). So I marked it as a blocker.
Michael V (not M) was seeing this on Friday - not sure if he logged a different bug. He had the problems view in a completely inconsistent state where it didn't reflect the errors in the classes of his workspace. Also, the status line at the top of the problems view did not always match the rest of the view... for example it would say "Filter matched 0 of xxxx" items", but there would be several problems in the table. There has been no change to the underlying marker infrastructure in 3.1 at all, so it looks like a problems view problem.
I think this is a dup of Bug 113582 but I am going to leave it open until we are sure.
This problem is still occurring for me in I20051025-0800. Either Bug 113582 is not related or it is not fixed.
Doug are you still seeing this (it has been 3 months)?
I have still seen this from time to time, though it is decidedly less frequent.
I have found a replicatable case 1) Import Ant Core from existing plug-ins 2) Open the class AntBuildfileContentDescriber 3) Remove the import that has the access warning About half of the time your markers will not update with the errors you need (and vice versa when you add it back).
I have released a fix for the problem below in build >20060203. I am not going to close this bug yet because Doug reported it before the code I just changed was written so there may be other issues as well.
*** Bug 128949 has been marked as a duplicate of this bug. ***
I found some more holes in the update story which I have fixed in build >20060402 by removing some flags, locks and unneccessary work (this will also give some performance improvements as well)
*** Bug 128951 has been marked as a duplicate of this bug. ***
Note that pre 20060303 the updates would have periodically have stopeed because of Bug 132838. We now have code that protects against this issue.
Is this fixed then? What's the plan for 3.2?
Yes. It was fixed in M6 with the fixes to update if there is no post update.
*** Bug 83168 has been marked as a duplicate of this bug. ***
*** Bug 101367 has been marked as a duplicate of this bug. ***
*** Bug 116623 has been marked as a duplicate of this bug. ***
*** Bug 127549 has been marked as a duplicate of this bug. ***
Verified in 20050425