[EMAIL PROTECTED] wrote:
I was still really wondering what the layers were that are
involved when sending a string from the command-line to Emacs
and how they alter parameters. CMD's 'readline'-like layer,
gnuclient's parameter parsing, Emacs' parsing, etc. and who
could be translating backslashes and requiring quoting ("I need
how many backslashes do I need to pass one in? Four?").
But if this %var:x=y% expansion will work I may not need to
worry.
Gnuclient just expands filenames before sending them to gnuserv.
Filenames must then be single arguments to gnuclient. Actually that is
not very good for the integration with the OS. You have to specifiy full
file names to -e when it works this way. I would prefer that when -e was
used that gnuclient first told gnuserv to change directory and then
evaluate the argument to -e.
On the other hand -- nothing stops me from writing this extension to
gnuclient/gnuserv. However there will soon be a successor to
gnuclient/gnuserv which is called emacsclient/server. (Unfortunately no
one has yet got this working for MS Windows). BTW - should not this
discussion by on the MS Windows port list?
