Community
Participate
Working Groups
When a BadLocationException is thrown in DocShare.java (somewhere in lines 432-451) getLocalizedMessage() is used to retrieve information about the exception. It returns a "null" message, which is a bit confusing. There should be a more "verbose" message :)
Changing target milestone and assigning to Mustafa. Mustafa this bug is WRT the message in the BadLocationException...but we also need to address the synchronization leading to a BadLocationException at all (as we saw at EclipseCon). If you want another bug for this, please create one.
This is serious. I suspect that with the introduction of the Sync API, something got broken/wasn't abstracted properly. I am looking into it right now and will try to either fix the Sync API/whatever else is the source of the error(s) or - in the worst case - try to revert to bare-bones Cola. The BadLocationException is just one symptom of the disease, which - in general terms - is broken sync. That is, even without the reported BadLocationException, the document states - with the current DocShare code - can diverge. After some initial testing, I have verified, that this can even happen *silently*, that is without any indication that something went wrong and documents are out-of-sync. The reason for me suspecting the Sync API is, that IIRC the Cola sync code wasn't changed since the Google demo last year. That's why I'm looking into the places where the synchronization code interfaces with other components of DocShare.
(In reply to comment #2) > This is serious. > > I suspect that with the introduction of the Sync API, something got > broken/wasn't abstracted properly. I am looking into it right now and will try > to either fix the Sync API/whatever else is the source of the error(s) or - in > the worst case - try to revert to bare-bones Cola. Unfortunately, this is not an option (reverting to bare-bones Cola) at this point. Just for reference, the impl of Cola used for the sync API is almost identical to the original Cola code...and is in org.eclipse.ecf.internal.sync.doc.cola.
Upping priority to critical, changing target milestone and title. Mustafa, please advise about how we can jointly isolate and make progress on this bug (you, Marcelo, myself and possibly others).
https://bugs.eclipse.org/bugs/show_bug.cgi?id=321228 might be a duplicate of this (is occuring in windows with php)
this bug is currently marked as critical, but we haven't seen any progress on it for almost two years...so I'm moving this down to major. I'm also adding 'helpwanted' to the keywords. If this bug can't be reproduced, or has already been fixed, then please update appropriately.
ping...Mustafa...as the keeper of this ECF code, what do you want to do with this bug? Can it be reproduced?
(In reply to comment #7) > ping...Mustafa...as the keeper of this ECF code, what do you want to do with > this bug? Can it be reproduced? I don't remember whether we got around to fix this at some point or not. I'll check out the code later today/tomorrow and try to reproduce this bug.
Please reopen if resources available to address.
