[gdal-dev] Large GeoJSONs and aborting file opening

[Even Rouault]

Simon,

I don't think that killing a thread is a safe practice in general. It 
would likely result in memory leaks and maybe in some other inconsistent 
state that could crash the whole process.  An interesting enhancement 
for such cases would be to be able to provide a progress / abort callback.

Even

Le 28/07/2021 à 21:22, Simon Eves a écrit :
> Dear All,
>
> I am aware that some improvements were made in the 2.3 timeframe with 
> regards to dealing with large GeoJSON files, although even in 3.2, 
> it's still very slow and memory hungry.
>
> Our system allows for aborting imports, but this only works reliably 
> once it has actually got to the stage of reading features from the 
> file. With the GeoJSON, it just sits in the GDALOpenEx call for ages.
>
> My question, therefore, is whether it might be practical to run the 
> GDALOpenEx in a separate thread with a future to return the resulting 
> handle, such that it could be monitored and killed if necessary?
>
> Mainly I would be concerned that killing the thread might trash some 
> global GDAL state that might then not be recoverable, or that the open 
> relies on some TLS for the process thread and therefore might not work 
> properly.
>
> We're going to try it anyway, but opinions welcomed, thanks!
>
> Simon
>
> -- 
> <http://www.omnisci.com/>
> 	
> Simon Eves
> Senior Graphics Engineer, Rendering Group
> 100 Montgomery St (5th Floor), San Francisco, CA 94104, USA
>
>
> 	
> Email: simon.eves at omnisci.com <mailto:simon.eves at omnisci.com> | Cell: 
> +1 (415) 902-1996
>
>
>
>
> _______________________________________________
> gdal-dev mailing list
> gdal-dev at lists.osgeo.org
> https://lists.osgeo.org/mailman/listinfo/gdal-dev

-- 
http://www.spatialys.com
My software is free, but my time generally not.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/gdal-dev/attachments/20210728/b7a90833/attachment.html>