#2638: runInteractiveCommand on Win32 leaks, produces inconsistent behavior
-------------------------------+--------------------------------------------
Reporter: sclv | Owner:
Type: bug | Status: new
Priority: normal | Milestone:
Component: libraries/process | Version: 6.8.3
Severity: normal | Resolution:
Keywords: | Difficulty: Unknown
Testcase: | Architecture: vax
Os: Windows |
-------------------------------+--------------------------------------------
Comment (by claus):
Replying to [comment:1 simonmar]:
> What happens in the "noread" case is that `runInteractiveCommand` spawns
a shell to run "./rcTest noread", and then you terminate the shell by
calling `terminateProcess`. Windows doesn't have the concept of process
groups, so it doesn't kill the child process.
I was surprised by this last comment, as I had seen the option "end
process tree" in Windows Task Manager (right click on a process). I
haven't found anything on "process trees" directly, but there seems to the
idea of a "job object" collecting groups of processes:
http://msdn.microsoft.com/en-us/library/ms684161(VS.85).aspx
That might not help in this particular case, if there's an intervening
process that doesn't play the game, but it might still be useful in
general when process groups are needed. It would be nice if everyone
wouldn't use different names for similar concepts..
--
Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/2638#comment:2>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler
_______________________________________________
Glasgow-haskell-bugs mailing list
[email protected]
http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
