On Fri, 2008-06-20 at 13:08 -0400, Richard Lowe wrote:
> >> That's clearly a bug in the python shim between the two, we shouldn't
> >> do that, and it will need to be fixed. (you'd be just as screwed if
> >> you did 'hg cstyle', or should have been...)
> >
> > I don't see the problem when running 'hg cstyle'.
>
> That surprises me, we should be filling that particular pipe either
> way, shouldn't we?
The different may lie in how the processes are plumbed together.
in the command-line case I'd suspect that the subprocess's stdout/stderr
goes directly to the user's terminal; in the server case, I'd expect it
would be the parent process which is capturing the output to embed in
the hg server protocol.
if there's any chance the output will be voluminous, your best bet is to
create a separate thread or process to pump data into the hooks while
the main process reads the hook output.
- Bill
