Menu
▾
▴
#3331 GCL vs. SIGPIPE
On my Ubuntu 16.04 box, when a broken pipe occurs a second time at GCL (or Maxima running over GCL), it is ignored: unless it is explicetely (re)allowed: (si:allow-signal 13); AFAIK it is not reallowed at Maxima, thus after having it once, it is not propagated anymore then.
It has a coupled issue that if a broken pipe occurs and is followed by (si:allow-signal 13) but not by another broken pipe (generally when the last allow-signal is not followed by broken pipe), the quit() function does not complete due to some (spurious) broken-pipe event. One has to run quit() again to actually end Maxima.
Am not sure whether Maxima should explicitely deal with it or whether it is intentionally left to users to manage it. It is not a big issue, more of a nuisance, unless one deals with it more. I had filed a ticket at GCL web; am not sure if they'll do anything with it though.
Thanks for the report. How does the broken pipe occur?
Hi, first of all, broken pipe does not occur at Maxima proper, Maxima is well.
It is that when you play more with outer stuff, it may occur; an archetypal case is working with gnuplot. Though the current (5.2) version of gnuplot is quite stable (at least according to my use, comparing to 5.0 version of it), thus it is not occurring within plot2d/plot3d for me now.
Not sure what to do here. I guess we need an example to help sort things out.
Greetings! GCL now restarts SIGPIPE, so this bug should be closed too.