Memory Leaks/Crash bugs...

[Szekeres Tamás]

Thank you for sparing no effort to discover theese sort of
problems.   You are not alone by making desktop applications
utilizing mapserver as a mapping engine. (Sometimes i'm
feeling alone with theese problems as well). I have found
similar issues with mapscript that caused heap corruptions
with some situations may ascribed to memory leakage. Calling
msCleanup at program termination still causes unexpected
dbgbreak in the debugger at the CRT code. The code compiled
with VS2005 especially suffers with this problem, and i have
open issues with the MONO runtime as yet.

Some of the problems like this could be dedicated to the
usage of different heaps (by using different versions of the
CRT code for example) and allocating and releasing the same
segment separately. This problem could easily be handled
during compilation i think.

By debugging the code frequently sometimes i run into
problems like
http://mapserver.gis.umn.edu/bugs/show_bug.cgi?id=1545
cannot be discovered easily at all. I think it is our turn
to make patch with proposed solutions since we naturally
have the environment to test it in action. I have planned to
fit in my activity flow to dig into theese issues more
thoroughly.

Actually it seems that the code was not developed primairly
to serve long running applications, but i'm optimistic with
the possibility to enhance by eliminating theese sort of
problems sooner or later.

Don't hesitate to file a bug if you have localized a problem
to be handled by changing the code even if you still does
not have a solution on it.


Best Regards,

Tamas Szekeres




Ned Harding <nharding at EXTENDTHEREACH.COM> írta:

> 1st - Just an FYI - I fixed a major crash bug in the label
cache when
> you have LOTS of objects.  I'll get Jani to review and
commit for me
> when he has some time.
> 
> 2nd - memory leaks
> 
> Again with LOTS (20,000+) of label items I am getting a
LOT of memory
> leaks with the following stack (This is MapServer 4.6 code
still)
>     s:\3rdparty\gfx\freetype\src\base\ftsystem.c (109):
ft_realloc
>     s:\3rdparty\gfx\freetype\src\base\ftutil.c (137):
FT_Realloc
>     s:\3rdparty\gfx\freetype\src\base\ftobjs.c (2356):
FT_CMap_New
>     s:\3rdparty\gfx\freetype\src\sfnt\ttcmap.c (2179):
> tt_face_build_cmaps
>     s:\3rdparty\gfx\freetype\src\sfnt\sfobjs.c (609):
sfnt_load_face
>     s:\3rdparty\gfx\freetype\src\truetype\ttobjs.c (209):
tt_face_init
>     s:\3rdparty\gfx\freetype\src\base\ftobjs.c (931):
open_face
>     s:\3rdparty\gfx\freetype\src\base\ftobjs.c (1635):
FT_Open_Face
>     s:\3rdparty\gfx\freetype\src\base\ftobjs.c (987):
FT_New_Face
>     s:\3rdparty\gfx\gd\gdft.c (473): fontFetch
>     s:\3rdparty\gfx\gd\gdcache.c (124): gdCacheGet
>     s:\3rdparty\gfx\gd\gdft.c (898): gdImageStringFTEx
>     s:\3rdparty\gfx\gd\gdft.c (804): gdImageStringFT
>     s:\3rdparty\gis\mapserver\maplabel.c (314): msGetLabelSize
>     s:\3rdparty\gis\mapserver\mapgd.c (2870):
msDrawLabelCacheGD
>     s:\3rdparty\gis\mapserver\mapdraw.c (1897):
msDrawLabelCache
>     s:\3rdparty\gis\mapserver\mapdraw.c (542): msDrawMap
> 
> Anyone know about this already?  If not I'll spend a
little time digging
> in to it.
> 
> I am guessing that we are the only people who have
embedded MapServer
> into desktop products and allowing users to throw
crazy/bad data at it?
> 
> Ned Harding
> Chief Technology Officer
> SRC - LLC
> 3825 Iris Ave Suite 150
> Boulder, CO 80303
> (303) 440-8896 x104
> nharding at extendthereach.com
> 
> SRC in Action:
> www.DemographicsNow.com
> 
> Join SRC at Extend 2006 - Disruptive Technology
> June 5-7, 2006 at the Hyatt Regency, Huntington Beach, CA
> For more details, click here: 
www.extendthereach.com/extend06/
> 

___________________________________________________________________________
[origo] klikkbank lakossági számlacsomag havi 199 Ft-ért, bankkártya éves díj nélkül! 
http://www.klikkbank.hu