<br><div class="gmail_quote">On Tue, Apr 8, 2008 at 3:19 AM, Jesus Cea <<a href="mailto:jcea@argo.es">jcea@argo.es</a>> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
I'm impressed. Very.<br>
<br>
I had considered a pyrex wrapper myself, months ago. When I became<br>
maintainer I thought that gradually refining current pybsddb would serve<br>
~ users better that throwing all the code and starting over, fresh. I<br>
didn't discard it, in any case.<br>
<br>
There are other issues, also. Originally pybsddb was started as a SWIG<br>
wrapper. Later it was rewritten in pure C. I don't know the details or<br>
the reason for that. Maybe Greg could comment about this.<br>
</blockquote><div><br>I know this point. Maybe that moment, the API is small, the wrapper including <br>core API and features writed in C is not so expensive, but more clear and <br>better controlled. <br><br>But now, the situation has changed, The API became huge, and We need <br>
full feature API in Python, along with the development of bdb.<br><br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><br>
Greg, are you out there?.<br>
<br>
Also, pybsddb is included in standard Python and no other module is<br>
writen in pyrex/SWIG. I don't know the reason, but I suspect they don't<br>
like to add dependencies, foreign bugs and so on. Maybe would be a good<br>
idea to just ask python-dev about this issue. I guess there is a real<br>
and strong reason for that policy. In fact, the oficial way to link to C<br>
code is "ctypes", that I find unuseable when the API requires<br>
heavyweight datatypes.<br>
</blockquote><div><br>pybsddb have less features and less codes, <br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><br>
| Is it a better idea to use Pyrex instead of raw C code to get more<br>
| productivity?<br>
<br>
My experience with pybsddb maintenance, with a couple of months already<br>
behind, is that current pybsddb code is useable and, in practice, I<br>
spend far more time documenting and thinking/writing the test cases than<br>
writing C. Those steps are needed too, even if the main core is Pyrex.<br>
</blockquote><div><br>Yes, the test cases and documents are also very useful when forking a new <br>Pyrex bdb wrapper.<br> <br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
But one of my motives to move to Python, a decade ago, was to forget<br>
about C programming. Life sucks :). Badly :)<br>
<br>
I would like to use your creative force and initiative to improve<br>
pybsddb. How are your C skills? :-).<br>
</blockquote><div><br>Today, there are too much things we need to do, but have ONLY 24 hours :-)<br>So, we MUST use tools to do what machine can do to improve our productivity.<br><br>If not neccessary, I will not write any C code. Because I can not write good C code<br>
rapidly.<br><br>Davies<br><br></div></div>