gno/kern/gno/TODO
1998-02-02 08:20:59 +00:00

89 lines
3.4 KiB
Plaintext

todo
====
- select() on pipes
- make console driver a loaded driver
- check drive full error code
- GetLang/SetLang
- access devices via /dev
- find out why tcp/ip takes so damn long to quit sometimes
- check p8 launching
- write up driver interface specs
revamp loaded driver interfaces so we can have stuff like /dev/audio
- how can we pass back correct error codes from SOCKrdwr to read() and write()
lib routines?
- create types, document diff between pid, kern pid, and kern table offset
baz.todo
========
- Rewrite the texttools to properly support texttools features
regardless of redirection. At the same time, implement a simple
buffering scheme for the texttools to greatly improve texttools
redirection. (must be tied into PGClose so we properly flush the
buffers all the time).
libs
====
- tc[gs]etattr, TIOCGETA, TIOCSETA
- getlogin
- sigemptyset, sigaddset, sigprocmask
reported bugs
=============
1/1/95 - init messages appearing on single line
done
====
10/27/97 wait() fixed to return -1, ECHILD if no children
11/5/96 select() should re-start after signal
should return EINTR after signal
with wait & *wait = 0 should basically not block
semaphore and KERNkill code re-worked to not spaz out GNO for
EINTR cases
4/30/96 commonSwait wasn't working right for EINTR when != procBLOCKED
3/31/96 libc: fixed %m and LOG_PID in vsprintf() (data bank problems)
libc: added gettimeofday()
libc: added setre[gu]id()
3/11/96 fixed stack deallocation problems in KERNkill and KERNexecve
3/7/96 fdCount++ was misplaced in dup2()
2/13/96 select() timeout now works (YEA!)
1/6/96 gs/os files working with select()
select() returns the proper value
? filed& filed& -> bogus process due to context switch during KERNkill()
socket close swait() problem fixed by removing some swait()
dependencies from socket close code
1/25/95 echo code in tty.asm wasn't calling signalIO like it should
have, select() on ptys seems to work now
1/15/95 2.0.6b4 released
1/12/95 ctool.asm: stuff added on 1/7/95 had incorrect entries (-1 doh!)
ps entry for init is now FUCK, FUCK, FUCKED! why? ACK! don't
link gno with lenviron installed!
1/11/95 2.0.6b3 released
1/10/95 receiveCommon - $0080.EOR.$FFFF was evaluating to $0000, thanx
a lot, mikey. This was causing the flags to get all fucked up
and bank 0 space was disappearing because UserShutDown() was
being used on forked memids instead of DisposeAll()
1/9/95 serial drivers were checking RR3 for SIGHUPing rather than
RR0
serial drivers do no SIGHUPing by default now - the special
condition interrupt enable and interrupt handler need to be
re-writen properly to handle different values for DHUP
1/7/95 more net stuff: shutdown(), getpeeraddr(), getsockaddr(), and
[gs]etsockopt() were added to the kern and libc
1/2/95 2.0.6b1 released
added version resource to kern
fixed InOutStart/End calling (I fuqed it up while playing with
making the console driver loadable)
libc: fixed strerror w/ network stuff
2.0.6b2 released
1/1/95 modem, printer drivers working with select()
correct orcalib from baz
welcome version 2.0.6b1
libc: added hosname stuff
libc: updated perror w/ network stuff
kernVersion returns $0206 now
12/?/94 null driver is now a loaded driver, and select() should work
right on it...
12/15/94 select() on sockets (ha! ha! bitch!)
12/12/94 Pass ioctl() on sockets directly to the usrreq routine
(do not pass go, do not collect $200)