lynx-dev Re: Slang - ncurses - TeraTerm

[Michael Warner]

On or about 26 Jul, 1999, T.E.Dickey
<address@hidden> wrote:

        [...]

> Teraterm is in the class (roughly half of all color terminals) that John
> Davis has shunned for several years.  If you clear the display, the current
> colors are not used to set the background color.  

OK, now I get it.  Before I went to a straight black background,
I always tweaked TeraTerm's background to match the slang color,
else it did, indeed, look pretty sloppy.

        [...]

> > I pipe an email from mutt to a script that filters the mail 
> > through a html'izing perl script to a temp file, and then starts 
> > lynx on the temp file.  The ncurses-built lynx I was using would 
> > flash the page up and immediately exit.  Using mutt's wait_key 
> > would hold the display on the screen, but that wasn't too useful 
> 
> iirc, the setbuf initialization in ncurses flushes the pipe - discards
> input - it's possible to code around this - by reading from a file
> descriptor rather than a stream, but Lynx wouldn't work anyway
> without some hacking (curses apps in general don't read from pipes)
> 
> otoh, if the Lynx slang configuration is doing something useful with a pipe,
> I'm not aware of that - the only use I see for pipe input in Lynx as it
> is now is some obscure logic in the command-line parsing.

I'll (try to) clarify:  Lynx isn't trying to read from the pipe,
as I understand it.  I mapped a key in mutt to:

"|${HOME}/bin/scripts/mail2html\n"

which sends the current email to the mail2html script:

#!/usr/local/bin/tcsh
formail -I "" $1 | ${HOME}/bin/scripts/txt2html.pl --mail > 
/tmp/warner/mutt-tmp.html
${HOME}/bin/lynx -blink -nopause -preparsed /tmp/warner/mutt-tmp.html
rm /tmp/warner/mutt-tmp.html

which strips mail headers, passes to a perl script that turns
URL's to anchors, and writes the result to a tmp file.  *Then*
lynx is started with the tmp file as its startfile.  I guess the
original pipe from mutt stays open until the mail2html script
finishes, but lynx isn't reading from it, right?

        [...]

> Lynx would have to read the pipe data (and do whatever it needs to),
> then reopen stdin from /dev/tty (or the equivalent on DOS or VMS).
> It could be done with curses by doing this before initializing curses.

Question 1 - Does (all or part of) this still apply in light of the above?
Question 2 - You're talking lynx source code changes, right?
             (If so, it's beyond my scope, I'm afraid)

-- 
Michael Warner
<address@hidden>