Date: 2011/10/17 - 11:22
Trac URL: http://trac.osgeo.org/mapserver/ticket/4048
Hello there, first of all thank you for this great project.
I've been working 5.6 and there was a bug and it still exists even in 188.8.131.52.
the bug is that there is an stackoverflowException raised when drawing the map which includes arabic labels with the following:
-when setting the encoding of the labelObj to CP1256
-when using IIS 6, 7 only,
not that :
IIS 5 or IIS 6 in IIS 5 compatibility mode or directly running from visual studio's Virtual server there is no problem and the map is drawn without any problems at all.
please let me know if you need any further details!
Date: 2011/10/17 - 11:38
forget to mention the framework, i tried 2.0, 3.0, 3.5 and 4.0
so it's not related to the framework.
it's either in mapscript.dll or what i 90% think is in LIBICONV
and btw, i also tried the TRANSLIT switch after the encoding like:
'''label.encoding = "CP1256//TRANSLIT";'''
it does not crash, but instead the characters are not related, you know arabic characters are connected to eachother.
Date: 2011/10/17 - 18:01
Could you provide a sample mapfile and test data to reproduce this problem?
Date: 2011/10/25 - 08:03
thank you tamas,
sorry for my late responce, i was kind of away from the internet.
Attached is the sample, but i havent included the Dll files or the zip file would have been 13MB!, so please copy all needed Dlls to the Bin folder.
Date: 2011/10/25 - 11:37
tamas, you might want to have a look at #4044 : fribidi is not threadsafe and may crash under multi-threaded use.
attachment http://trac.osgeo.org/mapserver/attachment/ticket/4048/Arabic Sample.zip :
Dear @Tamas, @szekerest
any news about this issue?
thank you very much.
I tried your sample but it didn't crash w3wp for me. Win7x64 IIS7, mapserver-5-6 (release-1600-x64 compiled from http://www.gisinternals.com/sdk/)
Dear @szekerest , Tamas
the issue is not yet fixed, any news about this?
Use thread lock around fribidi calls (#4048,#4044)
@ohdev I've added an (untested) fix in #4044 which I believe will fix your issue, albeit with a probably non-negligeable performance hit on multithreaded apps (there will be a thread lock for each rendered feature). When would you able to test this ?
Dear @tbonfort , i'm ready to test it, let me know where to download it.
thank you very much!
I don't do windows builds and can't help with that.
thanks @tbonfort, what do i need to do in order to test this or even build this if its something easy.
I have Vs2010 c++, if there would be a compiled version soon please let me know as it would be easier to test
have a nice day further.
@szekerest would you be able to provide a build of this branch to @ohdev so he can check this issue is fixed? we could then close this issue and incorporate in 6.2