Jamie:
> (...) I do know that under
> mod_perl Sablotfree() was definately being called, even in an error
> condition.
I see... I thought it was called from your own XS file. We'll check it.
> Weird thing is that setting it to zero caused a core dump when NOT running
> under mod_perl, and it couldn't access sablot.log (in Make test)
>
> Maybe the xs code needs to check the result and not call Sablotfree
> if it returned a non-zero, thats my guess anyway. (...)
Yes, definitely.
> Setting up a "message handler" would be nice, if it's convenient to do,
> (mainly just to avoid accessing a log file)
It's not much worse than just passing a pointer to a routine which then
gets all the messages. It would avoid writing to the log, but we'll put
a function controlling the logging into the coming release (0.35).
> I really like your package, and want to thank you very much for writing it.
> I look forward to using it.
Thanks too.
> You've probably already thought of this, but it's an idea that you
> might want to think about. (...)
> If there were some way to keep the parsed stylesheet in memory, you could
> apply several xml documents against it without having to re-parse the style
> sheet each time, "objects" or "handles" or something.
This would be cool! There should be something like this, I agree. One
could keep a cache of stylesheets, up to a limit on the cache size, I
presume. Unfortunately, the parsing now typically takes just a small
fraction of the total processing time, but when we optimize the
processor a bit, caching will become a very important issue.
Best,
Tom
