<br><div class="gmail_quote">On Sat, Jan 24, 2009 at 12:01 AM, Pete <span dir="ltr"><<a href="mailto:pfein@pobox.com">pfein@pobox.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
On Jan 23, 2009, at 2:04 AM, Drew Perttula wrote:<br>
<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div class="Ih2E3d">
Shannon -jj Behrens wrote:<br></div><br>
On another topic, does anyone have experience with the other dbm modules such as tokyo cabinet? I'm looking for something to use instead of berkeleydb since I'm sick of dealing with berkeleydb corruptions every several days. I can't manage to settle on a 'safe' config since I use multiple machines and distros. I'm ready to give up some write performance for robustness.<br>
</blockquote>
<br>
Yeah, we had various mysterious problems with berkdb a while back. Googling suggest "try this or that magic flag", at which point we switched to GDBM & have never looked back. To be honest, I've not heard of anyone who *hasn't* had serious problems with berkleydb in a non-trivial use case (best known example being Subversion).<br>
</blockquote></div><br>We(with one of my previous employers) had similar problems with berkdb when using the pybsddb module or the default module with Py. <br>We then latter wrote our own extensions modules using Boost:C++ and it worked like a charm. It was easy and we extended only those features which we needed.<br>
Also, we noticed that our module was faster than the pybsddb nmodule. Wish i could share the source!!!<br><br>-V-<br><a href="http://twitter.com/venkat83">http://twitter.com/venkat83</a><br>