Approximatrix Forums — Inelegant debugger exit leaves orphan process

Re: Inelegant debugger exit leaves orphan process

null@example.com (JohnWasilewski) — Mon, 05 Aug 2013 14:00:16 +0000

Excellent, Jeff, thank you. If it happens again with the current or future versions, I'll know now to put the skids under gdb.exe before annihilating my own creation!
---
J.

Re: Inelegant debugger exit leaves orphan process

null@example.com (jeff) — Mon, 05 Aug 2013 13:31:09 +0000

John,

The executable can't be killed because it is actually a child of gdb.exe, which is also probably left running. The latest change, available in the next hour or so, should fix this. It will continue to issue the elegant gdb shutdown commands, but, if gdb doesn't respond as many people are reporting, it automatically kills the process outright via a TeminateProcess call. It should be far more reliable.

Inelegant debugger exit leaves orphan process

null@example.com (JohnWasilewski) — Mon, 05 Aug 2013 12:08:26 +0000

A problem I've learn't to be wary of with the new SF2 debugger is that if I exit in an inelegant way, a Win Doze 'process' with the name of the executable I was debugging is left wandering the streets, and this prevents me from re-compiling the same executable after making changes and corrections to the code (as you do, when debugging).

In most cases, it's just a nuisance, because I have to run the Task Manager, find the process that's still there, blocking the compiler from SAVing, and put the process out of its misery.

Sometimes, however, it's a very, very BIG nuisance. Nothing but nothing I can do, save for a cold reboot, will get rid of the unwanted process. The Task Manager can't kill it, and nor can various other tools I keep in readiness for this kind of thing.

I have not yet managed to work out exactly what kinds of inelegant exits leave what kinds of troublesome debris behind, but I thought I'd just mention this straight away whilst I try to gather together a more lucid account of the problem. I daresay you're familiar with it already, Jeff. If it's possible to prevent it from happening, that would be great. If not, then some means of releasing a jammed-up process that SF2 or the debugger had initiated would be a great help.

All of the above used to happen with SF1.45 as well, but not nearly so much. Most of the time, the debugger put all its toys away when I sent it to bed, even if I shouted at it and refused to kiss it goodnight. Just occasionally, it left a process running.
---
John