Re: [PATCH CVS] log and colors and prettyprint

[Ilya Zakharevich <ilya@Math.Berkeley.EDU>]

On Sat, Oct 26, 2002 at 12:27:15PM +0200, Karim BELABAS wrote:
> >> I've only partially applied this patch. As Bill pointed out, escape sequences
> >> in logfile work nicely:
> >>
> >> * you can display them with cat, more, or less -r.

> > No.  Only if your TERM is the same, your background is the same etc.

...And only if you think that the screen peppered with symbols for
^A/^B is nice.

> Which is usually true.

Not here.  My setup in office (white background xterm) and two setups
at home (console-like windows with black and a darkblue backgrounds)
are mutiually incompatible for sharing screen escapes (similar to
light-gray on dark-gray problem).

> It is true that misc/gpflog could profitably be built in the logging
> mechanism, e.g
> 
>   log = 0: no log
>   log = 1: current behaviour
>   log & FILTER: strip all ANSI escapes + readline's <C-A>/<C-B> while logging

The logical order is

    log = 1: clean logfile
    log = 2: logfile with preserved escape sequences

> [ your patch only cancelled the ANSI escapes ]

Do not know what you are talking about.  The CVS version I have here

  GP/PARI CALCULATOR Version 2.2.5 (development CHANGES-1.571)

and the older version with my patch applied both produce logfiles
without any line noise.

So I'm content with the behaviour of 1.571.  If somebody wants to
implement log=2 (logfile with preserved escape sequences), feel free
to do it.  Since I do not care about log=2, I do not care whether
these logfiles have ^A/^B or not ;-).

Ilya