Re: some comments on htuml2txt.pl

[Christian Vogler <cvogler@gradient.cis.upenn.edu>]

Hi Golda,

thanks for the access to the commercial version. I'll try it when I get a
chance. Currently, I am still waiting for our web hosting company to give
us clearance for installing Webglimpse. Hopefully, this will happen
sometime this week.

> Here is the message, I suspect some of these things are intrinsic to
> glimpseindex and we can't affect them by a filter program. But the
> case-sensitivity should be doable...

I have not looked too closely at the e-mail yet, but here are my first
impressions:

The case-sensitivity problem really seems to be a Glimpse problem. I have
the suspicion that it does not use locales correctly when it compares
strings. 

At the moment I don't think that it will be possible to work around the
case-sensitivity problems by changing htuml2txt or htuml2txt.pl. If things
stay the way they are now, we get correct results with case sensitive
searches. If we change them around, we can get correct results with
case-insensitive searches, but I see no easy way to achieve both, because
the filter does not know whether the search is case-senitive or
case-insensitive.

Of course, Webglimpse could always create a file, according to whether the
search is case-sensitive, and the filter could check for this file, but
this hack is horrible to contemplate.

- Christian