Thanks for your elaborate answer. I didn't realize that uint64 was not
so standard (I work on Linux). Very good point. I will stick to signed
ints for now.

And yes, it is not only the free beer that I appreciate!

All the best,
Gabriel

On Tue, 2007-03-06 at 13:51 +0100, Francesc Altet wrote:

> Well, this turns out to be a limitation of the numexpr package. Internally, 
> numexpr uses a limited set of types for doing calculations, and unsigned ints 
> are always upcasted to the immediate signed int that can fit the information. 
> The problem here is that there is not a (standard) signed int that can be 
> used to keep the information of an unsigned int64. In addition to this, you 
> will probably know that Win32 doesn't define such an uint64 type, so 
> supporting this in numexpr in a general way is not going to be easy in the 
> future (unless Win64 or Vista have introduced the uint64 type, but still, we 
> need several years until Win64/Vista will completely replace Win32).
> 
> So, your best bet right now is to avoid UInt64 types if you can. If you 
> absolutely need UInt64, the only way for doing selections with this is using 
> regular python selections. For example, if your table has a 'colM' column 
> which is declared as an UInt64Col, then you can still filter its values with:
> 
> [row['colN'] for row in table if row['colM'] < X]
> 
> However, this approach will generally lead to slow speed (specially on Win32 
> platforms, where the values will be converted to long python types).
> 
> > All the best, and many thanks for making PyTables freely available,
> 
> Well, it's not only freely available, it's also meant to be *free* (as in 
> freedom ;) software.
> 


-------------------------------------------------------------------------
Take Surveys. Earn Cash. Influence the Future of IT
Join SourceForge.net's Techsay panel and you'll get the chance to share your
opinions on IT & business topics through brief surveys-and earn cash
http://www.techsay.com/default.php?page=join.php&p=sourceforge&CID=DEVDEV
_______________________________________________
Pytables-users mailing list
Pytables-users@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/pytables-users

Reply via email to