On 4/29/2011 2:22 AM, Dima Kogan wrote:

Hi Chris. I've been looking around a bit more and found a patch to the PP
documentation that describes both the RedoDimsCode syntax and the '=' in the
Pars that you mentioned:

http://www.mail-archive.com/[email protected]/msg00320.html

Thanks for the reference.  It has been merged into
PDL git on sourceforge.

This has not been merged, but perhaps should be. Also, I did a preliminary port
of another library to PDL via PP (https://github.com/dkogan/PDL-Triangle). This
required an output piddle whose size wasn't known at runtime. I used $RESIZE()
for that and it seems to work well. I can document $RESIZE(), as I understand
it, in the PP POD, but the source has a concerning warning:

I'll have to let the PP experts give you feedback on the
$RESIZE() or other options.  Looking at the code, I'm
not sure how threading could work when the dimensions
depend on each set of data.  Maybe it would be clearer
just to make the output piddles by hand using XS since
the results are basically two blob-pdls for each set
of input points.

"RESIZE USED: DO YOU KNOW WHAT YOU ARE DOING???\n"

Any insight to the pitfalls of $RESIZE()?

Looking at the sourcecode, it does appear that $RESIZE()
allocates a new piddle of the needed size.  However,
per my comment above, I'm not sure if using PP to auto-
generate the outputs makes sense as far as threading.

Cheers,
Chris

_______________________________________________
Perldl mailing list
[email protected]
http://mailman.jach.hawaii.edu/mailman/listinfo/perldl

Reply via email to