Hi Arian,
I'm aware that there is a bug in the shape recognizer that sometimes
(fairly rarely, but often enough that it comes up once every few hours
for me) creates inappropriate large shapes out of short lines. I hope to
fix this bug sometime -- but that part of the code is very finicky. Even
though I did come up on my own with the algorithm in a moment of
inspiration, and the recognition of individual line segments or circles
is a clean conceptual idea that works reliably, the part about detecting
when a stroke consists of several line segments or when several strokes
combine into a polygon is rather strange and I don't understand all of
the features, some desirable and others not. I've tried to analyze the
bug but it's not obvious where it comes from at all.
I would tend to agree that you likely ran into an extremely bad case of
this bug that caused a gigantic shape to be created. (for the memory
allocation to request 3.5 GB of RAM, I think the shape would have to be
several billion pixels tall). Perhaps it would make sense as a temporary
fix to not let the shape recognizer produce shapes that extend well
beyond the area containing the strokes that were recognized and abort
the shape recognition if that is what the algorithm produces (i.e., if I
can't fix the underlying bug, at least avoid the rare cases where it
causes a crash).
I'm sorry for the loss of your file!
Denis
On 12/11/2013 11:55 AM, Arian Sanusi wrote:
> Hi all,
>
> I had xournal crashing on me this morning. The document I was writing was
> new, so was completely lost. happened after about 14 min oft quite fast
> writing (notetaking during lecture). Might have been something like an
> Out-Of-Memory kill, the machine was heavily swapping before crash ('Heavy' as
> in system freezes for several seconds swapping to an SSD).
> Another thing I noticed: I usually have the shape recognizer enabled, and
> sometimes with short lines it recognizes huge shapes (10 pixels -> 200
> pixels, very roughly and heavily varying). I could imagine that an extremely
> huge shape was recognized, that just blew out of RAM?
> These are all information (I think) I have of the crash:
>
> dmesg:
> [36032.805679] traps: xournal[9685] trap int3 ip:7f9acdba6e8d sp:7fff18cf51b0
> error:0
>
> xsession-errors:
> (xournal:9685): GLib-ERROR **: /tmp/buildd/glib2.0-2.36.4/./glib/gmem.c:165:
> failed to allocate 3667587280 bytes
>
> this is xournal 0.4.7 from debian jessie on amd64.
>
> I don't think I can reproduce this bug. I use xournal exactly like this for
> several years for most note-taking for my studies, and I cannot remember
> another crash like this.
> If there is something else I can provide on this please let me know.
>
> Regards, Arian
--
Denis Auroux aur...@math.berkeley.edu
University of California, Berkeley Tel: 510-642-4367
Department of Mathematics Fax: 510-642-8204
817 Evans Hall # 3840
Berkeley, CA 94720-3840
------------------------------------------------------------------------------
Rapidly troubleshoot problems before they affect your business. Most IT
organizations don't have a clear picture of how application performance
affects their revenue. With AppDynamics, you get 100% visibility into your
Java,.NET, & PHP application. Start your 15-day FREE TRIAL of AppDynamics Pro!
http://pubads.g.doubleclick.net/gampad/clk?id=84349831&iu=/4140/ostg.clktrk
_______________________________________________
Xournal-devel mailing list
Xournal-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/xournal-devel
