home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.unix.bsd
- Path: sparky!uunet!usc!cs.utexas.edu!swrinde!emory!tackle!jc
- From: jc@tackle.uucp (John Capo)
- Subject: Minor XFree86 problems, long function key string bug in elvis
- Message-ID: <1992Dec30.005909.15079@tackle.uucp>
- Organization: Technology Atlanta Corporation
- Date: Wed, 30 Dec 1992 00:59:09 GMT
- Lines: 77
-
- I have XFree86 running and only have a few problems left to solve.
-
- 1. Auto-repeat is very slow. I can turn it on and off via command
- line switches or in Xconfig but I can't change the repeat or
- delay rate.
-
- 2. Num lock does not work at all.
-
- 3. I have no bell. I have tried various command line switches and
- xset options but nothing. Same hardware works fine with DOS and
- ISC running ISC.
-
- I have not seen these problems mentioned on the net.
-
- Another thing I have noticed is that my machine is on its knees when
- I am compiling and running X at the same time. It may take 20 seconds
- to expose a window. Same hardware with X386 and ISC only exibited this
- slowdown when compiling, running X, and my cron job that rebuilds the
- uucp paths file and uuwho database kicked in. I added a second disk
- with 16M of swap but no difference.
-
- No name 386/33, 8M RAM, 128K cache.
- Adaptec 1542B, 2 Maxtor 120MB, Archive 60MB tape, Julian's driver.
- Compuadd generic ET4000 VGA card.
- WD8000 ethernet card.
- All of the kernel patches in Terry's patchkit.
-
- Without X running the machine is very speedy, much better raw disk I/O
- performance than ISC with the same hardware. My only conclusion is
- the ISC SysV kernel makes better use of the relatively small memory I
- have. Any thoughts? Is there a kernel parameter that wuld help?
-
- Another mystery. If I hexdump the keyboard I get a different escape
- sequence for the arrow keys than I do when I run emacs or vi. This is
- in an xterm window. No difference from the console. The function code
- keys are always the same. What is the magic here?
-
- My function keys emit 5 characters and elvis didn't particulary like
- that. He would get 3 of them on one read and think that he had matched
- one of the mappings. I tried different timeouts but still crashed.
- Here's my fix.
-
- *** tio.c.orig Fri Dec 25 19:02:23 1992
- --- tio.c Fri Dec 25 20:03:04 1992
- ***************
- *** 591,606 ****
- --- 591,617 ----
- {
- /* No, it wouldn't. But check for partial match */
- if (!strncmp(map->rawin, &keybuf[next], cend - next))
- {
- count++;
- }
- }
- }
- +
- + #ifdef __386BSD__
- + /* If 1 is returned for a partial match the caller will call execmap()
- + * with a NULL MAP struct. Not good. Setting notimeout and keytime=50
- + * did not help. May be a select problem? Still this code should not
- + * return 1 if the match is not complete.
- + */
- + if (count == 1 && match == (MAP *)0)
- + return (2);
- + #endif
- +
- return count;
- }
-
-
- #ifndef NO_ABBR
- /* This function checks to see whether a word is an abbreviation. If it is,
- * then an appropriate number of backspoace characters is inserted into the
- * type-ahead buffer, followed by the expanded form of the abbreviation.
-
- --
- John Capo jc%tackle.uucp@mathcs.emory.edu gatech!vta!tackle!jc
-