[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: LYNX-DEV ac-0.81 doesn't like "x-euc-jp" in META

From: Klaus Weide
Subject: Re: LYNX-DEV ac-0.81 doesn't like "x-euc-jp" in META
Date: Thu, 16 Oct 1997 18:30:11 -0500 (CDT)

On Thu, 16 Oct 1997, Nelson Henry Eric wrote:

> FWIW, if I put "<META HTTP-EQUIV="Content-Type" CONTENT="text/html;
> CHARSET=x-euc-jp">" in a document with EUC encoding, Lynx breaks lines
> where I wouldn't expect it to.  If that is changed to "euc-jp", dropping
> the "x-", everything is fine.

This misbehaviour was introduced only a few snapshots ago, when I made
Lynx recognize "x-euc-jp" as meaning the same as "euc-jp" but failed to do
so consistently in all places where the charset is checked.  If you want
an immediate workaround, locate HText_setKcode in GridText.c and add a
line for "x-euc-jp" analogous to the other lines there.

> Yes, you ask why do you put "x-euc-jp"
> when it is wrong.  Well, Netscape renders "x-euc-jp" in the way the
> document was intended, but renders "euc-jp" in what I think is Russian-
> istic!

I have no idea what Russianistic Japanese looks like. :)
Anyway, this probably depends on what version of Netscape ***i*ator you
are using.

> As I mentioned before, it might be wise to ignore `CHARSET='
> in a meta tag when CJK is involved.  Compare, for example "http://163.
> 51.110.11/XEUCindex.html" and ""; which
> are the same except for the "x-" difference.

But, if you venture out of the circle of Japanese web sites and happen to
visit a document that specifies its charset correctly (let's assume
non-CJK, for simplicity), then you should get the benefit of that; i.e. 
the document should be interpreted according to its explicit charset, 
without you having to toggle '@' for a reload.  So CHARSET= shouldn't
be always ignored, even if CJK is 'ON', for this reason.  Whether this is
actually what is happening is up to you to test (and report); try the new
test/???88592.html files, they have an explicit charset in META.

By the way, for pages you make available from the server under your
control, consider testing also what happens if you specify the charset
"the right way", i.e. not in an embedded META tag but in the HTTP headers.
NCSA/1.5.2 should be able to do it, althogh I don't know the specifics.


; To UNSUBSCRIBE:  Send a mail message to address@hidden
;                  with "unsubscribe lynx-dev" (without the
;                  quotation marks) on a line by itself.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]