On Tue, 17 Jan 2006, Garrett Rooney wrote:
> On 1/17/06, Peter N. Lundblad <peter@famlundblad.se> wrote:
>
> This seems reasonable to me, but I still wish there was a way to
> return to the situation where hook scripts that exit success don't
> have to worry about this sort of thing. I imagine there are a number
> of scripts out there with this sort of problem, even our examples
> don't bother to close stderr, if we're going to require that we should
> at least fix the default templates.
Where in our default templates do we background any jobs?
>
> In any event, if we do go with this fix I think we should include some
> documentation on the correct way to correctly close stderr for various
> systems. Your example here covers unix shell scripts, but I'd
When you say "close stderr", I guess you mean redirect it, since that's
what my example does. This is important, since closing any of
stdin,stdout,stderr is a Bad Thing to do. That'll make the process output
stderr output (if any) to some random file that it might have opened
afterwards.
> specifically like to include something for windows batch files (not
> that I have any clue how to do that in a batch file, but I suspect
> people will ask for it).
>
AFAIK, it is the same on Windows
command 2> nul
But don't ask me how to start something in the background on that
platform...
I agree that we should announce this somewhere vissible for 1.3.1.
To address Philip's point about inheriting stdin, it seems good to
redirect stdin to the null device in that case in the callers and clarify
the documentation for svn_start_cmd.
Regards,
//Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Jan 17 23:45:25 2006