From: drew@kinglear.cs.colorado.edu (Drew Eckhardt) Subject: Re: MGR questions Date: Mon, 1 Jun 1992 00:40:18 GMT
In article <1992May30.172225.15062@daffy.cs.wisc.edu> jamest@cat24.cs.wisc.edu (James Thomas) writes:
>I've been playing around with MGR (the Sunday, May 24 version), and have
>some questions:
>
>1) If Bash (v 1.12) is the current shell, MGR fails to start it in windows
> it creates. If ash (the crummy default shell on the root disk) is the
> current shell, it is started up and I can run commands normally. Is
> this a known bug? (I'm running Linux 0.96a if this helps).
There's some wierdness (TM) in MGR, where the slave end of the
ptys aren't being opened. If you manually open the slave end,
you can run virtually anything -
ie
doshell /dev/ttyp0 /usr/local/bin/bash &
I've noticed the same behavior with tcsh too.
>2) I've copied the termcap info for MGR into my termcap file, but still
> certain keys fail to function. For example, backspace doesn't work
> in ash, and sometimes enter doesn't work in MicroEMACS.
"backspace" is not the backspace key (^H). It is "rub out". If
you to a stty erase to that character, it will have the same
behavior as ^H.
>3) Does the font support work yet?
>
For some definition of the word work. Some of the programs
in usr/mgr/demo/sh had bad escape sequences, and work once
this is changed.
I can do a font <N>, where <N> is one of the loaded fonts
(as set up in /usr/mgr/font/.mgrc).
>4) I notice that the demo programs use Sgtty, and thus cannot be compiled
> on Linux. What is the eventual solution? Rewriting the demos or writing
> a Sgtty emulator?
>
Writing the stty and sgtty syscalls / library functions. Rewriting
software to fit Linux is bad, because you do it whenver a new release
comes out. Replacing the missing system calls is what you want.
I'm working on stty / getty - already have TIOCSTART and TIOCSTOP
in there for termios, and have various other fixes to
.96 that will be out after I finish the pending SCSI driver
changes (these include support for disconnect / reconnect on the
Seagate lowlevel driver, meaning it doesn't just sit there
while accessing the disk, fixing the error handling code,
and nicer error messages.).
>Despite these problems, I rather like MGR and am glad that somebody is
>porting a window manager that is more reasonable in its demands on memory
>and disk space than X. (Not everyone has 16 MB of memory and 100 MB free
>disk space!)
mgr is pretty nice.