[Kst] branches/work/kst/portto4/kst/src/datasources/ascii

[Nicolas Brisset]

Hi Peter!

> SVN commit 1319572 by kuemmel:
> 
> read very big files not at once
GREAT! I can't wait to test this. Can you provide a Windows snapshot easily?
If I understand the code correctly, we try to buffer the whole file into memory, and if it fails due to missing RAM then we do it in chunks.
If that's correct, it means we don't even need to add an option in the GUI to "Don't buffer the file into RAM", as I was previously planning. Very nice...

Now, I hope we can solve the Purge bug (https://bugs.kde.org/show_bug.cgi?id=256153). I don't know if Barth has been looking into it yet. I started looking at it with the debugger here, but in my test case there are thousands of objects in the store and it doesn't make it easy to debug! It seems that the _used property is not set correctly on some objects, but I haven't figured out exactly which. Purge apparently goes through the list of objects in the store and just deletes those which are not used. Including automatic scalars, apparently?
I hope Barth will be able to help there as it's not trivial. The _initialxxxnum properties also have a strange behavior, but maybe it's just because I don't understand them well enough.

Once the Purge bug is solved, I'd vote for a 2.0.7 release as we'll have fixed a good number of annoying things. Time reading from ASCII can wait a few weeks more... and is probably going to take longer.

How do you see things?

Nicolas