Re: setopt and alias questions

[Sweth Chandramouli <sweth@xxxxxxxxxxxxxxxxxxxx>]

On Mon, Feb 08, 1999 at 12:55:00PM -0800, Bart Schaefer wrote:
> On Feb 8,  2:59pm, Sweth Chandramouli wrote:
> } Subject: Re: setopt and alias questions
> }
> } On Mon, Feb 08, 1999 at 11:44:03AM -0800, Bart Schaefer wrote:
> } > On Feb 8,  2:15pm, Sweth Chandramouli wrote:
> } > } 
> } > } for DIRNAME in ${fpath} ; do
> } > }    if [[ -x ${DIRNAME}/*(.x:t) ]] ; then
> } > }       autoload ${DIRNAME}/*(.x:t)
> } > }    fi;
> } > } done;
> } > 
> } > That loop will never autoload anything -- unless there happens to be a
> } > file in the current directory with the same name as one in ${DIRNAME}.
> } > Do you see why?
> } 	yes--because i need to start cut&pasting my functions rather
> } than typing them in.  that should be -n, not -x.
> 
> That won't work either (and my remark about "unless there happens to be a
> file in the current directory" is wrong as well).  Filename generation is
> not applied to the strings inside the [[ ... ]].  

	are you sure about this?  the above (with -n instead of -x) works
fine for me--or at least, the filename generation part does.  what i hadn't
realized until now is that the error i was trying to catch, about there not
being any matches for that glob, was being generated by the shell instead
of the autoload command; any dirs in fpath that don't contain executable
files, then, would cause the shell to return the same error when attempting
the glob in the [[ ... ]], which defeats the purpose of the entire loop.
	(globbing in the test operator _definitely_ does work, however.)
	could a flag be added to the list of glob qualifiers that says "if no
matches are found, quietly return nothing but a non-zero exit status"?  i
think "q" (for "quiet") is still free.

	-- sweth.

-- 
Sweth Chandramouli
IS Coordinator, The George Washington University
<sweth@xxxxxxx> / (202) 994 - 8521 (V) / (202) 994 - 0458 (F)
<a href="http://astaroth.nit.gwu.edu/~sweth/disc.html";>*</a>