Thanks...<br><br>In that case, I can try to convert to using lists<br>(as much as I hate to stop using dictionaries...<br>guess I can't have simplicity & speed :-(<br><br>For speed, where (and how) should one create these
<br>lists of unsigned long longs? Just use Numeric?<br><br>Regards,<br>Sam<br><br><div><span class="gmail_quote">On 3/14/07, <b class="gmail_sendername">Greg Ewing</b> <<a href="mailto:greg.ewing@canterbury.ac.nz" target="_blank" onclick="return top.js.OpenExtLink(window,event,this)">
greg.ewing@canterbury.ac.nz</a>> wrote:</span><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Sam Tannous wrote:<br>> From what<br>> I can gather, any access to a dictionary (whether created in python or pyrex<br>> code) slows everything down to pure python speeds.<br><br>That's correct. You can gain some speed by using
<br>various PyDict_* functions directly, but all the<br>keys and values still have to be Python objects,<br>so there's a limit to what you can achieve.<br><br>If you need a blazingly fast mapping from<br>integers to integers, a Python dictionary just
<br>isn't going to give it. You'll have to implement<br>your own C-level data structure.<br><br>--<br>Greg<br></blockquote></div><br>