Apache OpenOffice (AOO) Bugzilla – Issue 39389
OOo crashes when attempting to open data files named "sxw"
Last modified: 2013-08-07 14:41:36 UTC
Last night I wrote the attached sxw document, just a few pars, on my laptop using m60. Transferred it to the desktop, loaded it into m65. Instant, repeatable crash. I think that makes it P1
Created attachment 20725 [details] sxw file crashes m65 every time
Not fully true... crashes are P2. Would be P1 if ANY .sxw would crash OO.
MRU->ES: the .sxw does not look valid (eveon OO 1.1.x could not open this). Please have a look
Please check if Edit - Paste Special only crashes when using "Formatted Text (RTF)". Then it would be of issue 38569 and you could close it.
Sorry, forget my las comment! Wrong issue! ES->DVO: there is aparently "nothing" (just CRs) in this file. The file is corrupt. Ok, but we shouldn't crash anyway... do you have a clue?
$ unzip fragment\ of\ end\ of\ ct\ apocalyptic\ column.sxw Archive: fragment of end of ct apocalyptic column.sxw End-of-central-directory signature not found. Either this file is not a zipfile, or it constitutes one disk of a multi-part archive. In the latter case the central directory and zipfile comment will be found on the last disk(s) of this archive. unzip: cannot find zipfile directory in one of fragment of end of ct apocalyptic column.sxw or fragment of end of ct apocalyptic column.sxw.zip, and cannot find fragment of end of ct apocalyptic column.sxw.ZIP, period. $ file fragment\ of\ end\ of\ ct\ apocalyptic\ column.sxw fragment of end of ct apocalyptic column.sxw: data $ dd if=/dev/zero of=ddd.txt bs=1 count=7290 $ md5sum fragment\ of\ end\ of\ ct\ apocalyptic\ column.sxw ddd.txt 1772ad389169272c7654873c7913b112 fragment of end of ct apocalyptic column.sxw 1772ad389169272c7654873c7913b112 ddd.txt Conclusion: By all means not a sxw -> adjusting summary. setting keywords (regression since OOo 1.1.3 does not crash but gives the "ASCII-Filter" dialog) OOo 1.9m65 only crashes on the sxw, not when trying to open the identical ddd.txt, so it is fooled by the file-extension. Opening the ddd.txt gives the ASCII-Filter Options. Since it crashes on linux as well -> OS to ALL original summary: "repeatable crash on loading .sxw file"
Created attachment 20742 [details] Cause m65 to lockup X11
Created attachment 20743 [details] strace of soffice attempting to process testarticle
Is there anything I can do to help? ger
dvo: Apparently these are two different issues. andrewb's file: As cloph noticed, this file is just zeros. This shouldn't crash, but even without the crash it is not going to load 'properly'. There might be an additional bug that something went wrong during 'save', although us writing only zeros is a rather unusual symptom, so I'm tempted to put the blame elsewhere. Either way, one would need a way to reproduce that problem to fix it. dvo->andrewb: Can you reproduce the situation where you save a document and it comes up all zero? grsingleton's file: This file contains hyperlinked graphics. The problem with TCP/IP networks is that they cannot give failure notices, so the only way to react to certain classes of errors is to wait for time-outs. I suspect this is what happens here. After some delay, it loads fine over here. The last statement in the strace indicates that is attempts to read from a file handle, which would support this. dvo->grsingleton: Please load the file and just wait for a few minutes to see whether it comes up eventually. Check whether the document, when it has loaded, contains graphics that can't be loaded. Also, the strace thingy tells me you are tech savvy enough to examine the reason for the time-out: Please do me a favor and determine whether the time-out problem is on our side, or elsewhere, e.g. server doesn't respond, proxy settings wrong, URL wrong, etc. (On the matter of time-outs: There's been discussion on whether we can do something better than time-outs or whether the time-out values should be changed. I'm not really sure whether the current state is optimal, but I surely will not open that can of worms. I suspect dev@openoffice.org would be a suitable forum.) Also, merry christmas and a happy new year to everyone. I'm going on vacation tonight, so I won't be able to work on this issue in the next two weeks.
Can do. If this is indeed timeouts then I have a problem with OOo tying up the entire desktop while waiting. Should we make an new issue?
I think it should be a new issue, as the two problems do not seem related. If office ties up the entire desktop, that would indeed sound like a bug. That's not what I have seen so far, but admittedly I tested neither the exact version you used nor the platform.
Created new issue for X11 hanging on network timeout http://www.openoffice.org/issues/show_bug.cgi?id=39538
Well, all I can say is that it wasn't corrupt when I saved it:-) I will have another look at the original file which was copied over from the laptop to the main machine when I synchronised the directory. It has never happened to me before that a file is corrupted in that way when being copied. So I suspect there was something wrong in the way m60 saved the file. But I can't repeat that. I will try, when my daughter allows me the use of the laptop, to reproduce that odd save. But I don't have very high hopes.
There's nothing wrong with your document. I added one that caused a problem similar to what you described. I have subsequently opened a new issue that addresses what I found and which is different from your experience. Happy holiday
dvo->andrewb: No news so far, so I assume it's not reproducible. Or maybe your daughter is just rather strong willed. :) I'll close the issue for now. I'll keep an eye open if other users report similar problems, but so far nothing of the sort has emerged. Andrew, if you run into the same thing again, or have additional information, feel free to re-open or re-submit the issue. Thanks for taking the time!
I'm happy to leave it like this. It hasn't happened again. Actually, there have been a lot of one-off crashes with the last couple of builds, and I am not sure what the prodecure should be. I tend to report any crash I have not seen before and can't find in IZ. But then I worry whether I'm making work for QA if they can't be reproduced. I think I'll ask on the QA list
Closed.