[geos-devel] Geos - multithreading and error handling

[Mateusz Loskot]

Sheng Liang (SH/CBC) wrote:
> Hi: Why did you say that geos is not safe for multi-threading? Do you
>  have any experience to support your point? Or the multi-thread safe 
> is not a goal of geos development? I am interested in this point 
> also.

I'm saying that from my experience in working with GEOS code, C++ API only.
I don't see any multi-threading/thread-safety techniques provided in
GEOS code.
There is no mutual exclusion, no indirection in raw pointer usage,
no guarantee of atomic operations in access to global/static data,
no usage of thread synchronization techniques, etc.

This tells me there is no thread-safety in GEOS.
Certainly, I may be wrong and there may be something I've not learned,
so I'm interested in deeper discussion too.

Cheers
-- 
Mateusz Loskot
http://mateusz.loskot.net