[mapguide-internals] RE: PATCH: Raster stability fixes, ticket#462

[Frank Warmerdam]

Haris Kurtagic wrote:
> Yes, I am aware of that, I write that in my first email. But, still it
> is semi fix. Depends when you reach pool size.
> 
> GDAL is not thread safe ( when I looked at source code more then year
> ago it was not). According to Frank with 1.6 multithreading is yet to be
> confirmed/tested, I hope Frank will comment this.

Folks,

It is intended that read-only access should be thread safe in GDAL as long
as each thread is using a distinct GDALDataset object (not sharing a single
one).  However, it is difficult to *prove* threadsafety in C++ applications
and it is likely that there are some holes in it in GDAL.

I would like to see the "big lock around GDAL" be something that can be
easily turned off for those willing to take a bit of risk, and do some
testing for their use case and want to get maximum performance.  However,
I'm a bit nervous about shipping MapGuide and the FDO GDAL provider in
a way that completely trusts GDAL's thread safety.

Best regards,

-- 
---------------------------------------+--------------------------------------
I set the clouds in motion - turn up   | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush    | Geospatial Programmer for Rent