brianw@hpausla.aso.hp.com (Brian Wallis) (05/07/90)
I'm posting this for a friend, Could replies be posted here and I'll collect them and forward them on. Thanks, brian wallis... Details.... ---- A vax (3300) running Ultrix 3.0, x11r4, xfig as contained in that distribution (direct from mit). Standard ultrix C compiler (hmm maybe I'll build gcc). Server is an NCD-19, doesn't seem to have a version number built in.... release notes say version 2.0.0. Problem, xfig dies in random ways.... Most common: Bad Request (seemingly random major opcode) next: Bad length in PolyText8 (maybe just an instance of above) occasionally: It gets an XIOError, broken pipe most of the time (all the time?...maybe). This happens when you do lots of text input, eg. get into text entry mode and hold your finger on a key, wait 'til the buffer fills then go do it again (and again until it breaks). Nick seemed to be able to get it to generate floating exceptions as well (when messing with latex-constrained vectors) but it doesn't do that for anyone else. Besides that's the sort of error you can find with a debugger (as long as it's not dbx which core dumps trying to load xfig, gdb works ok), the server receiving trash is a little more difficult to find. I seem to be able to get it to hang sometimes. It's down in select(2) waiting for input and doesn't seem to get anything back from the server. Oh..., I'm running twm from the r4 distribution. Everything else (from the mit distribution, not contrib) seems to run fine so I don't think that I could blame r4. I tried building the libraries without the optimizer but that didn't make any difference (welll, it was worth a try). I think the version from r3 did much the same things... I forget. The libs where built by saying...... make World >& world.out make DESTDIR=<somewhere-with-enough-disk-space> install make-a-bunch-of-links-out-of-/usr-point-at-it I seem to be able to get the same sorts of errors out of Vem, the graphics editor out of the Berkley CAD tools but I haven't messed with that very much. I guess it could be the server (i'm trying to get a new release of it's code) but I would have thought that other things would break if that where the case and it goes wrong on a VaxStation3100 running whatever server Dec supply with ultrix 3.0. Can't think of anything else to say about it.... got any thoughts? -------
envbvs@epb2.lbl.gov (Brian V. Smith) (05/08/90)
In article <3150002@hpausla.aso.hp.com>, brianw@hpausla.aso.hp.com (Brian Wallis) writes: |> |> I'm posting this for a friend, Could replies be posted here and I'll collect |> them and forward them on. |> |> Thanks, brian wallis... |> |> Details.... |> ---- |> A vax (3300) running Ultrix 3.0, x11r4, xfig as contained in that |> distribution (direct from mit). Standard ultrix C compiler (hmm maybe |> I'll build gcc). Server is an NCD-19, doesn't seem to have a version |> number built in.... release notes say version 2.0.0. |> |> Problem, xfig dies in random ways.... |> Most common: Bad Request (seemingly random major opcode) |> next: Bad length in PolyText8 (maybe just an instance of above) |> occasionally: It gets an XIOError, broken pipe most of the |> time (all the time?...maybe). |> |> This happens when you do lots of text input, eg. get into text entry |> mode and hold your finger on a key, wait 'til the buffer fills then |> go do it again (and again until it breaks). |> The version of xfig from the MIT distribution (1.4.3, I believe) used setitimer() to blink the text cursor. That is an inherently unstable method of doing anything in X that uses the X library, as the library is not re-entrant. You should pick up xfig-2.0 from expo.lcs.mit.edu or comp.sources.x. It has several enhancements and bug fixes. _____________________________________ Brian V. Smith (bvsmith@lbl.gov) Lawrence Berkeley Laboratory I don't speak for LBL, these non-opinions are all mine.