dcw@doc.ic.ac.uk (Duncan C White) (09/08/89)
Hi everyone,
I have a few requests and thoughts:
1). Several times in the last couple of months, I have downloaded new
sources, taken them home and then found that they require some other
library routines/include files/kernel enhancements to run.
I do find this a right pain - couldn't people, when posting source,
AT LEAST say up front what non-standard bits and pieces are used.
In some cases, it might even be worth including the extra stuff
so that we can all get to run the new software without having to
send lots of messages taking up space in the newsgroup.
(Although this would make the postings larger).
2). ST-specific. All the time-setting routines I've seen go past are
for battery backed clock boards. I have a bog standard 1040ST.
When I use TOS (gag, splutter) I have a little program I wrote a
while back which reads the time out of the keyboard clock, which,
although not battery backed, survives resets. This means I only
have to set the time once, when I switch the machine on. From then
on, after a reset, I can accept the defaults the time program gives
me.
I wonder: has anyone done this under Minix? Sounds trivial, but of
course, I don't have access to the XBIOS traps. The question is,
therefore, how do I read and set the keyboard clock from Minix?
This breaks down to: how do I send and receive packets to the IKBD,
especially considering that the keyboard driver is presumably grabbing
these packets, and might be unhappy about a user process grabbing
them instead.
Would I need to extend the keyboard driver to understand new ioctls,
in much the same way as the mouse support package does?
Any hints? ( I haven't, ahem, actually got around to recompiling my
kernel yet, so I'm definitely NOT a kernel hacker yet..)
3). Mainly ST-specific. Has anyone attempted to use the ST's ability
to support multiple screens (setscreen() in XBIOS under TOS) to
implement a simple 'virtual terminal' scheme? I have something
similar to that provided on Xenix in mind, where certain keyboard
sequences (ALT-Fn, I think) would switch to different virtual
terminals. Each virtual terminal supports a separate login.
It seems to me that this would do all I ever need out of windows,
while being enormously simpler than a windowing system. (A couple
of keyboard sequences to cut and paste text from one virtual screen
to another would complete my happiness).
Has anyone thought of such a thing? How difficult might it be?
4). Errm, does anyone have putenv() in their library? I downloaded a
program called "env" (by Terrence Holm, I think) which attempts to
modify the environment for running other processes. Unfortunately,
the ST library appears not to have this.
5). When I set up SVC the other week, I found that doing a co of any old
version didn't work. (Only checking the current version out worked!)
co said 'Checking out revision 1 of eek' and then gave a 'fix' error
message to the effect that the diffs-file contained a 'cat <<EOF' line.
(Which of course, it shouldn't have!!)
Recall that the archive format is:
cat <<EOF >{file}
{version 2 code.}
EOF
if test {version_requested} -gt 1 then exit 0 fi
cat <<EOF >diff_file
{diffs between version 2 & 1.}
EOF
When I looked at the source, it was apparent that, when co saw the
'if test' line, it did the test, and then started producing the diff
file. Clearly, it should have discarded the 'cat <<EOF >diff_file'
line. Purely and simply: it didn't.
I haven't seen a patch posted - did others work this out for
themselves, or is no-one using SVC, or has no-one ever extracted
an older version?
Duncan.
[ Reply to: dcw@doc.ic.ac.uk or ...!ukc!icdoc!dcw ]
----------------------------------------------------------------------------
Duncan White, | "But Holmes, that means Sir Charles painted
Dept. Of Computing, | the hound before his death."
Imperial College, | "Yes, Watson: he is unlikely to have painted
London SW7 | it AFTER his death!"
England. | Dudley Moore/Peter Cook