From: Lachlan Andrew <lha@us...> - 2004-04-25 13:25:43
When I profile the code, I get quite different results from Joe,
presumably because I don't use an external parser, and my host is
1. Could anyone confirm that gethostbyname is really as expensive as
Joe's profile suggests? If so, I'll write a cache for it. The
profile looks a bit suspect there, because gethostbyname is reported
as only being called a handful of times...
2. On my system, about 50% of the time is spent in HTML::parse(). It
looks ripe for optimisation. In particular, does anyone know why two
passes are made through the document? The first just seems to strip
comments/noindex and decode SGML tags. If I optimise this, the most
efficient way would be assuming 8 bit characters (include UTF-8).
Are we still planning to make ht://Dig unicode compliant? If so, do
we plan to use wide characters or UTF-8?
ht://Dig developer DownUnder (http://www.htdig.org)
Get latest updates about Open Source Projects, Conferences and News.