leo@aai.com (04/12/91)
b2676870@rick.cs.ubc.ca (wilson w ma) writes: >I encountered a very similar problem with that particular message but I >noticed something else when I put in a #define HAVE_SELECT. At that point, >Emacs had no problems coming up but none of the shells would work properly. >If I ran the TeX-shell, it would run TeX but the shell did not terminate TeX >completely and I would have to kill the shell before running TeX again. I >think I have resolved the problem, but I would like to know if you have >encountered any similar problems, what you did to resolve them and why those >problems occurred (I personally fiddled around with the PTY stuff). I have >not had the time to research the source enough to formulate the proper reasons >why it works now. Perhaps you have been enlightened more than I have ;) I had some problems with shells too. It turns out that they were caused by entries in my .cshrc file that were inappropriate for subshells. I added if ( $?prompt == 0 ) exit to the top of .cshrc (just after setting the path) and my subshells work fine now. I did nothing special with pty's except for enabling the max number in sysadm. There are still some problems remaining... Although the shells function properly, there seems to be a problem with the output from the subshell getting flushed back to emacs. When I'm running a make I often don't see any output after the make command is echoed until it's finished. If I hit any key in the buffer, the screen is updated and I can then see the intermediate output. I included xmenu support (Control,Shift,middle mouse button). There are a number of menu entries, like "list all keystroke commands", that cause emacs AND the parent window it was detached from to crash. Has anyone solved these problems? Are they specific to ISC, or do you see the same thing on SCO, ESIX, and Microport? Thanks for any info! -- Leo leo@aai.com leo%aai@uunet.uu.net ...uunet!aai!leo
kayvan@satyr.UUCP (Kayvan Sylvan) (04/13/91)
In article <1991Apr11.191131.27136@aai.com> leo@aai.com writes: Has anyone solved these problems? Are they specific to ISC, or do you see the same thing on SCO, ESIX, and Microport? I saw and fixed these problems in ESIX a while back. The problem is that the Emacs select() function is not enough of a select() emulation to work right for X windows (also, at least in ESIX's case, the system supplied select() has some problems with pipes and ptys). Adding the following to config.h solved this problem: #define select emacs_select And *don't* define HAVE_SELECT. You also need to make sure that BROKEN_TIOCGETC and BROKEN_FIONREAD are defined (at least in ESIX's case). Experiment with these and let me know. ---Kayvan -- | Kayvan Sylvan Consulting/Training 879 Lewiston Drive | | Sylvan Associates Unix and Mac Programming San Jose, CA 95136 | | "Think Globally, GNU tools, GUIs, Databases (408) 978-1407 | | Act Locally." Unify User's Group Librarian apple!satyr!kayvan |