User-Agent: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-GB; rv:1.9.0.11) Gecko/2009060214 Firefox/3.0.11 Build Identifier: 3.0b3 (can't get version as crashes - poss 12Apr or poss current) I was running a build that I thought was the latest - though on inspection I can't find any files newer than 12Apr so its possible the nightly updates have been failing? It hit the bug where parts of the UI appear to work, but the message window doesn't update - It couldn't be quit from the File menu so I quit from the ToolBar (did not have to Force Quit) When I started shredder again it crashed, and CrashReporter crashed. I tried downloading latest 3.1a nightly which also crashed I tried (from terminal window) a debug version (from 21 june), which crashed - see attached output from terminal window I deleted all *.msf files and tried debug - it failed, but in a different way - attached output from terminal window Reproducible: Always As said on other bugs, crashes are bad .... failure to restart and 100% loss of data should be considered a blocking bug. I'm on skype now "mitra_earth" if anyone wants to talk me through tests.

One more observation - recently TB had got really slow / unresponsive at times, and looking at Activity window it appeared to be reindexing every folder, and all msf files had new times, even those on folders not used recently.

In trying to fix it ... I found that ... If I deleted all the accounts (Library/Thunderbird/profiles/xxx/Mail/* it would start ok I could then replace the accounts one by one till I found the bad one, I could then start without that account, and replace the files a few at a time. The weird thing is that I eventually replaced ALL the data, i.e. even that in the Folder I indentified as causing the problem. This was seen once before, and used as a pointer that the date on some of these files (e.g. the date on the folder v. the date on the file) caused the problem where the ICC profile became unreadable, when really all the data is there and probably fine. Could I suggest fixing this is made a blocking bug ! *I* got all my data back by copying folders around, any inexperienced user hitting this problem is going to suffer loss of ALL their email (and I've caused similar corruption on more than one occasion now - see Bug 478595, so I don't think its going to be rare)

(In reply to comment #1) > One more observation - recently TB had got really slow / unresponsive at times, > and looking at Activity window it appeared to be reindexing every folder, and > all msf files had new times, even those on folders not used recently. you have Gloda indexing turned on?

If you mean "Enable Global Search & Indexer" in Advanced Preferences, then yes. - Mitra

(In reply to comment #6) > If you mean "Enable Global Search & Indexer" in Advanced Preferences, then yes. then comment 1 is a known issue related to gloda

(In reply to comment #0) > I was running a build that I thought was the latest - though on inspection I > can't find any files newer than 12Apr so its possible the nightly updates have > been failing? ? that's a long time to not notice nightly updates not working > It hit the bug where parts of the UI appear to work, but the message window > doesn't update - please start with a fresh install to an empty directory. And, given your ongoing crash problems, suggest turning off automatic update so that a) you always know what version you are running b) so that you reduce/control introducing changes, which makes diagnosis more difficult. Updating once or twice a week is probably sufficient. > It couldn't be quit from the File menu so I quit from the ToolBar (did not have > to Force Quit) > When I started shredder again it crashed, Is there any reason to believe this isn't just bug 478595? > and CrashReporter crashed. see also Bug 461702 - Crash Reporter crashed after submitting report > I tried downloading latest 3.1a nightly which also crashed > I tried (from terminal window) a debug version (from 21 june), which crashed - > see attached output from terminal window > I deleted all *.msf files and tried debug - it failed, but in a different way > - attached output from terminal window I don't see how "unable to read ICC Profile" is related. I don't see this being caused by Thunderbird. Do you have some java+Mac problem going on? Why did you open this new bug rather than post the terminal window in bug 478595?

Hi Wayne - there seem to be at least two, and possibly as many as four bugs here. 1: Why TB gets stuck in its UI 2: Why it corrupts data when Quit from the Toolbar 3: Why TB fails to start with the corrupt data 4: Why Crash Reporter crashes reporting the problem. If all that gets addressed is #1 or #2 then future bugs will continue to turn from minor problems (just restart TB) to major problems (unrecoverable data loss) While #3 & #4 are most like the same, #1 & #2 are most likely different from bug 478595 so I posted it as a new bug and refered to it in bug 478595. I don't have a java+mac problem that I'm aware of , everything else works fine - that error message "unable to read ICC Profile" has been previously suggested as a problem with CrashReporter. (which is part of TB I presume).

Mitra, how is the the restart behavior leading to "unable to read ICC Profile" with current nightly? When did you last see problem?

Hi Wayne, sorry but following up on a bug four months after the report isn't likely to catch generate a usefull memory :-) Having said that ... "unable to read ICC profile" is the error message on the last line of the first attachment. I don't remember seeing this problem recently - data corrupting crashes are pretty rare thank goodness - but due to the catastrophic nature for the user (loss of entire message base) I would suggest worth prioritising when they do occur, especially when - as in this case - the failure to recover from them was repeatable - but not any more as I don't retain the relevant files if noone seems interested in the bug !

I had seen ICC profile a couple years go, spring 2007 or 2006 iirc. Never seen it again until now. I researched it again but the notes are on another computer. ICC has to do with color management. So I would expect this to somehow be related to a mangled stack and some other software installed on your computer. As for now, closing incomplete