[GRASS-dev] vector large file support

[Markus Metz]

Glynn Clements wrote:
> Markus Metz wrote:
>
>   
>>> It seems to work, at least I got past the segfault when breaking 
>>> polygons, but I still have to check spatial index dumps.
>>>       
>> diff spidx.dump.orig spidx.dump.patched >spidx.diff
>> gives me an empty spidx.diff, patched spatial index is identical to 
>> original spatial index
>>
>> Waiting for Glynn's comment, then I would commit.
>>     
>
> What sort of comment were you expecting?
>   
Hmm, not really sure. Some assurance to submit the patch? You have more 
coding experience, you came up with a much more elegant solution than I 
did. I still feel uncomfortable with SVN, although I know that changes 
can be reverted. This segfault in the rtree lib doesn't happen often, 
and I don't know if I should submit that patch also to develbranch_6 and 
releasebranch_6_4. For now I have submitted to trunk only.

I started the whole business with vector LFS and the rtree bug, but that 
could be madness on my side. I would still like to get LFS for the 
vector libs, but I realized that other issues need to be sorted out 
first. The next issues on my list are the dglib cache bug and speeding 
up topology building (particularly areas/isles/centroids). Probably 
related to topology building speed, the category index could do with 
balanced binary search trees instead of sorted arrays. The speed gain 
for handling the category index could be similar to what I did for 
r.watershed.
I can not do all that alone, and highly appreciated your comments, so I 
would dive into the code and then try to come up with specific questions 
to the list when (not if, when) I get stuck, hoping for similarly 
helpful comments like for the rtree bug.