Re: init.sh changes pushed

[Eric Blake]

According to Jim Meyering on 2/16/2010 6:07 AM:
>>> +      *) feb_file_=$(echo $feb_file_ | sed "s,^$feb_dir_/,,;"'s/\.exe$//')
>>
>> How do we go about fixing this?  Should init.sh be given some smarts (like
>> autoconf-generated files) to re-exec the calling script using a saner
>> shell?
> 
> I like this idea.
> Encapsulating the smarts in the test framework itself
> would be more maintainable/reusable than what coreutils does.

Next question - how to learn the running script.

$ echo 'echo $0' > f2
$ cat <<\EOF > f1
> #!/bin/sh
> echo $0
> . ./f2
> EOF
$ chmod +x ./f1
$ ./f1
./f1
./f1
$ bash ./f1
./f1
./f1
$ zsh ./f1
./f1
./f2
$ ln -s `which zsh` sh
$ ./sh ./f1
./f1
./f1

Good: bash, dash, Solaris /bin/sh, and ksh all appear to preserve $0 into
a sourced script.  Meanwhile, zsh botches things when invoked as zsh, but
works when invoked as sh.  Since all of our test scripts use #!/bin/sh, if
the script was truly started via the she-bang, we should be fine even if
zsh is /bin/sh.  Any counter-examples out there of some other shell that
fails to print "./f1" on the second line using this example?

-- 
Don't work too hard, make some time for fun as well!

Eric Blake             address@hidden

signature.asc
Description: OpenPGP digital signature