[sword-devel] Comming soon: new improved sword searching

Lamar Owen sword-devel@crosswire.org
Mon, 9 Sep 2002 11:44:55 -0400


On Sunday 08 September 2002 04:12 pm, Chris Little wrote:
> FWIW, we need to upgrade our regexp engine.  The current one (from GNU)

> Perl Regexp fixes both of these problems.  We can use it under the
> Artistic License & it's got nice UTF-8 support built in.  I nominate
> switching to this as a .5 priority. :)

One distinct advantage to the Perl engine is that it is hacked on quite a bit, 
and has been optimized considerably, since Perl relies on regex support as a 
staple.

As to using regexes, I use them more than just occassionally -- but they are 
definitely a programmer's tool.  But there is no more powerful and expressive 
way of searching for some sets of words and phrases than a properly crafted 
regex.  Of course, if the existing one is DFA, and the replacement is NFA, 
then some regexes may break in subtle ways.... See the O'Reilly regex book 
for details on how deterministic finite automatons and non-deterministic 
finite automatons differ from the point of view of crafting regexes.
-- 
Lamar Owen
WGCR Internet Radio
1 Peter 4:11