[darcs-users] Strange bug in xdarcs.el (1/3)

[Rohan]

Trent W. Buck <trentbuck <at> gmail.com> writes:

> 
> On Sun, Nov 01, 2009 at 08:47:42AM +0000, Robin Green wrote:
> > Most likely this is a bug in emacs 23.1, then. I don't think it could
> > be a failure-to-flush-output bug in darcs, because emacs uses a
> > psuedoterminal to communicate with subprocesses (at least in this
> > case, anyway), and so such a bug would have been evident when using
> > darcs directly.
> 
> Without further information, I'd like to suspect haskeline here.  Can
> you tell xdarcs to use stdio instead of allocating a pty?  That ought
> to convince Darcs that it's talking to a script (not a human).
> 
> You may also want to check how vc-darcs.el and darcsum.el do things,
> since AFAIK they both still work, though I haven't checked lately...

Yes, turning off pty use fixes this.  In:

  (defun darcs-do-interactive-command (root-dir patch-responses &rest options)

wrap the "start-process" call like so:

    (let ((process-connection-type nil))
      (setq process (apply 'start-process cmd-line "*darcs output*" "darcs"
options))

cheers
Rohan