Jason Rumney wrote:
Jason Rumney wrote:
We already have w32-system-shells, which is the inverse of the
variable you propose (except it is missing "cmdproxy.exe", I am not
sure if there is a good reason for that).
In fact we have a function w32-shell-dos-semantics for precisely this
purpose. So a suitable patch might be as below. The only thing left is
to silence the compiler warning about w32-shell-dos-semantics not
being defined on other platforms. Is (defun w32-shell-dos-semantics)
the right way to do that?
Very nice. Seems like the way to go. Together with Kim's changes (which
I believe adds calls to shell-quote-argument here) this should fix the
problem.
But I have two questions:
- Is there not a unix style shell for DOS too?
- What rules should be used for applying shell-quote-argument? I mean
where should it be applied? Should it not be applied as late as possible
when forming the command line to send to the shell? Do we do that now?
BTW, wonder why this was not implemented before? It looks like the
function w32-shell-dos-semantics were added already 1998, eight years ago.
_______________________________________________
emacs-pretest-bug mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/emacs-pretest-bug
