On Sun, Nov 28, 2010 at 04:20:21PM -0500, Joseph Xu wrote: > > > On 11/28/2010 06:22 AM, Aur??lien Aptel wrote: > > On Sat, Nov 27, 2010 at 12:35 AM, Joseph Xu <joseph...@gmail.com> wrote: > >> came up with. It relies on the shell that executes st to parse the > >> arguments, so you can't run a command like st -e "touch arst", you have > >> to run st -e touch arst. This also means you can't have any st arguments > >> after the -e because they'll all be included in the command to run. > > > > This breaks argument parsing. Why would you want to do that? > > Josh's solution is better imo. I've pushed (a slight variation of) it to > > tip. > > Not sure what you mean by breaking argument parsing. I think it's the > simplest way to parse a command with complex quoted arguments correctly > without having to run an extra shell process. >
Agree. I prefer this patch to mine. Mainly due to a bug(?) in dmenu_run where everything is escaped before being executed by the shell. So, for example, I can't run `st -e "tmux a"` from dmenu_run, since it parses it as st -e \"tmux a\" instead. The xterm-like -e behavior circumvents this issue.
