Re: [ppc patch] fix grub_ofconsole_cls on Old World

[Hollis Blanchard]

On Sep 14, 2004, at 10:34 AM, Marco Gerards wrote:

> Hollis Blanchard <address@hidden> writes:
>
> >         (grub_ofconsole_cls): instead of ^L, use an ANSI escape sequence and
> >         grub_gotoxy(0,0) manually.
>
> What I did is correct according to IEEE 1275, IIRC.

Remember that there is no guarantee any particular IEEE1275 
implementation has followed all of the IEEE1275 spec. *Particularly* on 
Old World Macintosh.

> I assume you are
> using a terminal emulator that connects to your mac using a serial
> line.  So in this case we are fixing GRUB to behave like what your
> terminal emulator expects, instead of the ANSI subset used by IEEE
> 1275.  Am I correct?

Hrm, it seems you are correct.

When I enter ^L in a bash shell and log the output, I see that it has 
really emitted "^[[H^[[J". When I print ^L from GRUB, I see that it has 
really emitted ^L.

'screen' is the only convenient terminal emulator for me, and when it 
receives ^L it apparently doesn't do anything (it does correctly 
interpret the ANSI sequence above). However, when I use minicom, the 
screen clears correctly with ^L. I'm willing to bet that minicom also 
correctly interprets the ANSI sequence.

> It would be better to have something like a simple termcap database in
> GRUB.  It could be very simple, I think.
[snip]
> Is this idea I have crazy and is there another way to do this?

You're scaring me. :) It would be nice to allow screen to work, but at 
the cost of implementing termcap in a bootloader? I think it could be 
put on the "known bugs" list and addressed later.

Isn't it worth it to require ANSI support for now, which increases the 
number of working terminal emulators? We already depend on ANSI for 
moving the cursor at least. If you do end up implementing a termcap 
(which I do think is extreme), all ANSI dependencies could be fixed at 
once.

-Hollis