Richard Stanton <stan...@haas.berkeley.edu> wrote:
> > > 2) If not, where is the code that interprets %s? I can try stepping
> > > through it and see what I can find.
> > >
> >
> > It happens in org.el:org-open-file, specifically this piece of the code (I
> > think):
> >
> > ,----
> > | (while (string-match "%s" cmd)
> > | (setq cmd (replace-match
> > | (save-match-data
> > | (shell-quote-argument
> > | (convert-standard-filename file)))
> > | t t cmd)))
> > `----
>
> I did some tracing through, and have found where things go wrong.
>
> At the start of the code listed above,
>
> cmd = "c:/progra~1/sumatrapdf/sumatrapdf.exe %s"
> file = "c:/dropbox/org/personal.pdf"
>
> (convert-standard-filename file) returns
> "C:/dropbox/org/personal.pdf"
>
> So far so good...
>
> However,
>
> (shell-quote-argument (convert-standard-filename file)) returns
> "c\\:dropbox/org/personal.pdf"
>
> That's where the problem occurs. After this block of code is completed, cmd
> takes the value
> "c:/progra~1/sumatrapdf/sumatrapdf.exe c\\:dropbox/org/personal.pdf"
>
> This command causes sumatrapdf to squawk with an error. It's the "\\:" that
> causes the problem. If, after this block of code executes, I manually switch
> the "\\:" back to ":/", so cmd goes back to
>
> "c:/progra~1/sumatrapdf/sumatrapdf.exe c:/dropbox/org/personal.pdf"
>
> and then press "c" to continue, the PDF file pops up just fine. Could we just
> drop the call to shell-quote-argument? It seems like everything would work
> fine without this call.
>
Can you trace the execution of shell-quote-argument? In particular, what is the
value of system-type and what result do you get when you evaluate the form
(w32-shell-dos-semantics)
?
It seems that it falls through to the default case and I'm not sure
it should.
Nick
