Re: compile trouble

[Ben Pfaff]

John Darrington <address@hidden> writes:

> This seems to be a problem with non bash shells expanding
> string comparisons wrong.

Can you explain?  Where do you see evidence of that?

> ie we need to write "test x$foo = xyes" instead of 
> "test $foo = yes" .

test "x$foo" = xyes is safer.  There are two issues:

        - Double quotes protect $foo from shell word splitting,
          so that if $foo has a space in it you still end up with
          just one word.

        - The reason for the leading "x" is different.  It is
          explained in the Autoconf manual:

             Similarly, Posix says that both `test "STRING1" =
             "STRING2"' and `test "STRING1" != "STRING2"' work
             for any pairs of strings, but in practice this is
             not true for troublesome strings that look like
             operators or parentheses, or that begin with `-'.

             It is best to protect such strings with a leading
             `X', e.g., `test "XSTRING" != X' rather than `test
             -n "STRING"' or `test !  "STRING"'.

In some circumstances you can omit both.  A fair amount of
Autoconf code does that, because it knows that $foo must already
be set to some safe string (e.g. to exactly "yes" or "no").

> There are places (both in our configure.ac and in gnulib - in 
> particular the include_next module) where this is not happening.
> I sent a patch to the gnulib list partially fixing the
> problem, but unfortunately got a somewhat aggressive response
> - "We cant apply patches that make things less reliable".

The only examples I see of "test" in gnulib/m4/include_next.m4
are the case where the variable should already be set to a safe
string.  Can you point out an example that you think is wrong?

> I don't know why they think it would do that. I've not 
> found a sufficient length of time where I've been in the
> correct mood to compose a reply yet.

I know how that goes, sometimes (but not usually with that
mailing list).

I don't recall that discussion there, do you have a pointer?

Thanks,

Ben.
-- 
Ben Pfaff 
http://benpfaff.org