[tlug] Mozilla - comment and questions

[Jim Breen <jwb@example.com>]

G'day all,

How do TLUGers feels about Mozilla, and the state it has reached with
1.0? I have been poking around with it for a while, and am pretty
impressed. It does so many things much better than Netscape 4.xy that I
am close to making it my Browser-Of-Choice. There is, however, at least
one aspect that makes it hard for me to leave Netscape. More below.

Anyway, first a question. I want Mozilla to be able to speak XIM
protocol with kinput2. To do this I need to launch it with the
appropriate locale variables set. At present I can do this using a local
"mozlaunch" script:

	#!/bin/tcsh
	setenv LC_ALL=ja_JP   (possibly this is not needed)
	setenv LC_CTYPE=ja_JP
	/usr/bin/mozilla

and it works fine.

Now what I really want to do is launch it from a menu or panel button,
but I can't get the environment variables set. I have tried setting up
Gnome menu items pointing at the above script, but Mozilla starts up
not talking to kinput2, from which I conclude that that the variable is
not set. On the chance that the Gnome menu used sh I set it to 
"LC_CTYPE=ja_JP /usr/bin/mozilla" which works fine in sh, but that
didn't work either. Does anyone have a suggestion on how to make this 
happen via a Gnome menu (this is on RH 7.1 and 7.2)?

Second, a big gripe about the way Mozilla handles Javascript and
Japanese text.

As some of you will know, I have a server that generates "bookmarklets"
for WWWJDIC. These are Javascript snippets that sit behind button on the
Toolbar an enable slices of text to be cut off zapped off to WWWJDIC (I
also have buttons for driving Goo and Google in Japanese mode.) The
Javascript has to be specialised according to browser and platform - IE
5+ sends out text in UTF8, Netscape 4.x sends out EUC on *n*x palforms
and Shit_JIS on Whinedoze platforms, so I add a code to the URL that
tells WWWJDIC what to expect.

Mozilla, just to be different, sends out the text in whatever the page's
coding is. So If I'm looking at a EUC-coded page, that's what it sends,
etc. etc. A real pain in the arse, as if the string is short it is very
hard to detect the coding reliably.

Does anyone have any brilliant ideas on this? If nothing else works I'll
work on tuning a code-detection scheme, but it would be much better if I
could pluck a variable value from somewhere.

Until this is fixed, I am stuck with Netscape 4.x for most of my reading
of Japanese pages.

Cheers

Jim

-- 
Jim Breen  [j.breen@example.com  http://www.csse.monash.edu.au/~jwb/]
Computer Science & Software Engineering,                Tel: +61 3 9905 3298
P.O Box 26, Monash University,                          Fax: +61 3 9905 5146
Clayton VIC 3800, Australia      ジム・ブリーン@モナシュ大学