hal@manta.mel.dit.csiro.au (Harold A. Miller) (10/04/90)
I am new to MH, but already have an annoying, intermittent core-dumping problem with mh-6.7. I hope somebody knows what I need to do! Environment: Sun SPARC, SunOS 4.1, running in a window on a Macintosh IIcx (NCSA Telnet 2.3.2). One user, with a .mh_profile of *reasonable* size and content, can coredump repl (and I think comp, and a couple others, but it seems he uses repl the most). He can reply to the same message, in the same environment, using exactly the same keystrokes, repeatedly, and only crash it once in a while. We're still trying to pin it down. I've gone through source until it's coming out my ears. I've found a couple little anomalies (possibly leaving an 'fp' opened, etc) but nothing that seems likely to either dump core, nor even likely to occur. Only thing I don't like is that the 'putenv()' routine doesn't free memory--just dereferences (reassigns) the pointer, each iteration. The core dump is, according to dbx on those few occasions we've been able to trap, always occurring at a getenv() call in sendit(), in whatnowsbr.c. The user has a signature file, has annotate and inplace set, and (I think) no relevant altmsg. I don't think the problem is what getenv() is called for, but perhaps that the environment space has been trashed by putenv() previously. Anyone seen anything like this? Anyone have any ideas? My user is getting rather impatient with my explanations! Thanks much. Hal -- |Hal Miller, DIT, CSIRO, | Networking Environments Project | |55 Barry St, Carlton, | (TEL) +61 3 347 8644 (FAX) +61 3 347 8987 | |VIC 3053, Australia | Internet:hal@mel.dit.csiro.au | -- -- |Hal Miller, DIT, CSIRO, | Networking Environments Project | |55 Barry St, Carlton, | (TEL) +61 3 347 8644 (FAX) +61 3 347 8987 | |VIC 3053, Australia | Internet:hal@mel.dit.csiro.au |