Ciao Gabriele!

At 6:48 AM -0400 4/16/99, Gabriele Bartolini wrote:
>I've noticed that Ht://Dig uses HTTP/1.0 to retrieve documents. What
>happens if we use HTTP/1.1 instead of it, by any chance? There would be any
>compatibility problems with servers that don't support HTTP/1.1? And
>overall, are there any meaningful (for htdig, obviusly) differences between
>them?

For the last year or so, my goal has been to get HTTP/1.1 going. Fully
compliant code would be wonderful for ht://Dig because HTTP/1.1 allows
byte-ranges (great for update digs) and persistent connections. If ht://Dig
implemented all the features of HTTP/1.1, there would be a pretty
significant speed improvement, especially for those of us indexing a small
number of servers. One connection could be made for all files, eliminating
all the TCP overhead on each connection. The W3C has some great articles on
benchmarks, not surprisingly using a minimal web spider.

The problem is I haven't had the time or the motivation to work out
HTTP/1.1. As for compatibility, if we implement HTTP/1.1 correctly, we're
supposed to drop back to HTTP/1.0 if the server doesn't recognize 1.1.

Randy Winch submitted some patches to let ht://Dig use the libghttp library
(part of the GNOME system), but initial performance wasn't great and I
didn't have the time to figure out what was going on. It seemed like
persistent connections either weren't working, or were dying out because
htdig wasn't making the connections fast enough as the database grew. Thus
the focus on improving database speed. :-)

>The second question is: why doesn't Ht://Dig uses a date class to store and
>manage in a Object Oriented way the date information?

This is a wonderful idea. We could put this in htlib/ and have all the code
in one place. It should probably be able to translate between many of the
UNIX formats like tm, time_t and of course, text.

-Geoff


------------------------------------
To unsubscribe from the htdig3-dev mailing list, send a message to
[EMAIL PROTECTED] containing the single word "unsubscribe" in
the SUBJECT of the message.

Reply via email to