On 1/20/19 11:44 PM, Bill Allombert wrote: > On Fri, Jan 18, 2019 at 03:28:04PM +0100, Tobias Hansen wrote: >> Package: src:gap >> Version: 4r10p0-6 >> Severity: wishlist >> Tags: sid buster patch >> >> Hi Bill, >> >> a version of the remaining patch that sagemath applies to gap was >> merged upstream a while ago [1] (and tagged for backporting to gap >> 4.10). embray indicated in [2] that the patch is important for >> non-interactive use of gap so I wanted to ask if you would add it to >> the Debian package now that it is merged. > Hello Tobias, > "non-interactive use of gap" is quite vague. Does it affect sagemath or > not, and how ? We are past the point where we have time to try out new patches > before the release. This is not a new GAP problem that need to be fixed > in a hurry. > > Cheers,
Hi, yes it affects sagemath. From the PR: "A more realistic use case is one where GAP is started as a subprocess with stdout and stderr redirected to pipes, and then one of those pipes is closed." This can happen in sagemath if sage is closed while waiting for a gap command to return. Without the patch it can leave behind a gap process which is in an infinite loop. I haven't yet looked deeper into the problem with gap-io. Best, Tobias