thread safety and OGR

[Frank Warmerdam]

Tamas Szekeres wrote:
> Frank,
> 
> I'm +0 on this issue since it is a symptomatic treatment but we should
> also discover the particular reason of the issue. We should also
> aspire to support the independent execution of the multiple threads as
> much as we can.
> As far as I remember the GetFieldAsString problem of OGR have been
> treated as of RFC-6 as a "side effect".
 >
> We might want to ask the client for testing with the latest OGR. It
> would be also practical to see whether your temporary fix really
> solves the client's problem or not. If yes I would be interested in
> tracking down the root of the problem at the OGR level.

Tamas,

My client is apparently using C# mapscript, and will be providing me
with details to reproduce the problem and verify it's correction.  I'll
also be preparing some sort of "multi-threaded heavy load" tester to help
verify correction operation.

Co-incidentally, I also have a contract to implement some degree of
thread safety within OGR (for GDAL/OGR 1.5.0) though that doesn't
offer any guarantee that every driver will be threadsafe.  So some work on
the root of the problem is also underway though for that there are always
"conditions" (only verified for some drivers, only safe if no two threads
use the same OGRLayer, etc).

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    | President OSGeo, http://osgeo.org