paul@cacilj.UUCP (Paul Close) (09/09/88)
I've just finished getting X11 R2 up and (mostly) running on an Apollo 4000
running the brand new SR 10.0 (FCS [First Customer Ship]). Well, MOST of
it works! I have not built an Apollo X before, so I don't know of any "tricks"
that may be required. Here are my problems:
Building:
o the new cpp is up to making Makefiles, so I did. Make sure and
define "-A nansi" since cpp doesn't do token concatenation like
ANSI says it should (a##b -> ab). Apollo, are you out there?
o server/ddx/apollo/ap_text.h is missing a #endif! I don't know
why this wasn't noticed earlier! server/ddx/apollo/apc/apctext.h
has the same problem. Also, do an obj2coff on apcfont.bin.
o After the fixes, lib/X/XConnDis.c references <netinet/tcp.h>
which doesn't exist. I just commented this out. Will this
cause problems?
That's all I remember. After that, most things worked. The notables that do
work are xterm, xclock, xinit (Xapollo), xconq (yea!), xmille, and xedit. On
to the problems:
o in xconq, a call was made to usleep which bombed for some reason.
I replaced it with a call to select.
o The cursors (i.e. the "X" cursor, the "I" xterm cursor, etc) in
monochrome are not correct. The "X" doesn't have a white
border, the "I" is actually just a "X" with the edges trimmed,
and the "L" uwm cursor is actually an arrow! This is only
in monochrome--color works just fine in this respect.
o programs that use SendXYImage hang up in _XSend doing an infinite
loop through WritevToServer (a/k/a writev) and _XWaitForWritable.
what happens is writev returns -1 with errno=EWOULDBLOCK, which
causes a trip through _XWaitForWritable which immediately says
"go ahead", and writev returns the same thing, causing an infinite
loop. What's up here? Are Apollo's sockets not working?!?
The program that doesn't work is xphoon (and setroot -bitmap).
Note--this is a rather hefty sized bitmap (960x848), but it
*should* work. xsetroot -bitmap hangs in the same way on really
large bitmaps.
o if the above bug doesn't bite xsetroot -bitmap, on bitmaps like
"flagdown" (48x48) or "woman" (75x75), the tiling doesn't work
correctly. The tiles run all the way across the top of my screen,
wiping out my xterm! The "top" happens to be the top of my other
xterm window. Sounds like an algorithm bug....
o for reasons unknown, the keyboard and mouse will freeze up and I
have to reboot. The machine still seems to be running, but I
can't do anything without a keyboard or mouse! Once it happened
after running fine all night, another time during testing of
xphoon (see above). Is this the server jamming up, or what?
o not really a bug, but who decided to put "meta(POP)" so far away
from "shift([v])" and "control([^])"?!? Try doing a "M-shift" as
required for uwm (do people actually use this server?!?). I have
a "type-3" keyboard, and I'd like to use it, thank you.
I can't proceed from here. Any help would definitely be appreciated!
I hope X11R3 works better than this!
Thanks for listening,
--
Paul Close paul@cacilj.CTS.COM ...!{uunet, ucsd, crash}!cacilj!paul
Shaw's Principle:
Build a system that even a fool can use, and only a fool will want to use it.