Re: LYNX-DEV Re: chartrans

[Leonid Pauzner]

> > Is it possible to add ASSUME_CHARSET to O)ptions screen
> > since changing lynx.cfg may be restricted by sysadmin
> > and command line flags not very convenient
> > (especially if you want to change default during the lynx session) ?
>
> It is possible and probably desirable, but I probably won't do it now for
> the old reason: overcrowding of O)ptions screen...  If/when we do to a
> FORM-based Options screen, OR someone integrates the FORM-based solution


I think one more line in O)ptions still exists...


> > There is a strange behavior found in ac.49:
> > I set display charset to cp866
> > and look at local document which has META...=windows-1251:
> > tuggling \ (html source) leave russian text the same,
> > while for the document with META...=koi8-r
> > tuggling \ give "7-bit approximation" !
> >


I mean there is an obvious bug in certain chartrans table.
I check "=" for each document above and see:
charset=windows-1251
charset=windows-1251 (assumed)    [for \ source], and that's OK
charset=koi8-r
charset=koi8-r (assumed)       [for \ source]  but that is wrong:
                        the real "7bit approx" appears on the screen instead!!!



> > Maybe this is corrected in newer versions, but they are not available
> > for me as DOS binaries.
>
> No, I don't think there have been relevant changes.
> I think you have run into something else, not a difference between
> windows-1251 and koi8-r.
>
> If the "... ;charset=windows-windows-1251" is (also) found in the HTTP
> header, it is always used.  The ']' key should show you the HTTP headers.
>
> If a "charset" is only in a META tag, Lynx can only know about it if the
> HTML is parsed and the META is interpreted.  Normally Lynx would forget
> what it knows about a document when you switch to SOURCE with '\', since
> it has to be reloaded, and then the normal ASSUME_* would be in effect.
>
> As a partial workaround, there is some "remember the charset from the last
> time" when '\' is used, but it is not complete.  It does not work if Lynx
> _totally_ forgets about a loaded text before reloading, and whether that
> happens depends on various things in the wwwlib mechanism which keeps
> track of documents and links between them, and which didn't change from
> previous versions.  Approximately, if there are links from other loaded
> documents to the current one, then the "total forgetting" does not happen.
> So if you have just followed a link to a HTML text, and then type '\',
> Lynx should use the same charset; if you went to the current doc with 'g'
> and there are no links to it from where you were before, or if it is the
> startup (first) document, Lynx doesn't remember the previous charset.
>
> Is this acceptable?
>
> As a workaround, you could try to go to the text whose source you want to
> see through a link; adding it to a bookmark file (maybe temporarily) and
> then going from there should have this effect, also going through the
> 'V'isited Links Page (but not the History Page), also, if it is a local
> file, going through a directory listing (but not if local dired is in
> effect since there is some extra expiring going on - I think this caveat
> doesn't apply for Lynx386 since dired is not compiled in (?)).
>
>
>    Klaus
>



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