Re: ./init.sh: line 139: `test-acos': not a valid identifier

[Eric Blake]

According to Ralf Wildenhues on 2/18/2010 1:57 PM:
>> Thanks for the offer, Ralf, but I feel pretty strongly
>> that adding $EXEEXT as a suffix to every invocation would
>> constitute "too much" pollution, and for what?  To enable
>> mingw-like systems to run tests using a portable shell.
> 
> IIUC then it's not for MinGW.  It's only for cross setups where you have
> an emulator but the host system doesn't emulate $EXEEXT-interpolation
> for you.  Right?  (On MinGW, .exe is appended implicitly in all
> interesting situations.)

Yes, I agree - this is mainly for wine (about the only system out there
where there is an emulator, but the emulator does not do implicit $EXEEXT
handling).  But wine runs on Linux, where we can be reasonably assured of
having a better-than-POSIX shell, and thus rely on POSIX extensions like
'-' in function names, as part of the workaround specific to wine.  The
workaround isn't needed on cygwin or mingw, but doesn't hurt, and those
platforms also come with a guarantee of bash.

The only other platform that init.sh is worried about is lack of $() in
the default /bin/sh of Solaris; there, the workaround does NOT have to
find bash; it is merely enough to find ksh (aka /usr/xpg4/bin/sh), so that
the rest of any script that included init.sh can assume the common base of
POSIX features.

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

Eric Blake             address@hidden

signature.asc
Description: OpenPGP digital signature