[Libreoffice] [BUG] Hide system pkg-config, to avoid finding stray freetype.

[James C]

Yes, we can't protect against an arbitrarily broken system.

I'm not a good systems administrator.  I don't know how to do "please
consider hiding it" without breaking something else.

I think pango is faulty; it needs to depend on how cairo was
configured, but it is testing how the system is configured.

But that's above my pay-grade.  I like this change because it's within
the part of the world that's currently open to me to possibly
(depending on this conversation) change.

On Thu, Dec 22, 2011 at 11:59 PM, Tor Lillqvist <tml at iki.fi> wrote:
>> This is a protection-from-contaminated-system
>
> But in general, a system can be contaminamed in arbitrary ways.
> Should/can we really protect against arbitrary, unknown, ways in which
> a system might have been changed by "helpful" 3rd-party software or
> misguided sysadmins/users to not correspond to a normal installation
> of the OS in question? No, we can't.
>
> What we should do, IMHO, is to check in our own configure.in if there
> is a pkg-config in PATH on a system where one is not expected to be
> present (only Mac OS X, I guess?), and in that case emit a warning.
>
> But wait, we already do that!
>
> if test $_os = Darwin; then
>    AC_MSG_CHECKING([for bogus pkg-config])
>    if test -n "$PKG_CONFIG"; then
>        if test "$PKG_CONFIG" = /usr/bin/pkg-config && ls -l
> /usr/bin/pkg-config | grep -q Mono.framework; then
>            AC_MSG_RESULT([yes, from Mono])
>        else
>            AC_MSG_RESULT([yes, from unknown origin])
>        fi
>        AC_MSG_WARN([This might have unexpected consequences, please
> consider hiding $PKG_CONFIG])
>        echo "Having a $PKG_CONFIG might have unexpected consequences,
> please consider hiding it" >>warn
>    else
>        AC_MSG_RESULT([no])
>    fi
> fi
>
> --tml