Peter Eisentraut <peter.eisentraut@2ndquadrant.com> writes:
> On 7/31/17 15:38, Tom Lane wrote:
>> Really? That seems pretty broken, independently of how many variables
>> are affected. But the ones you'd be most likely to do that with are
>> using AC_PATH_PROG already, I think. Having lesser-used program variables
>> behave inconsistently doesn't seem like much of a win.
> Well, if we're fiddling around here, I would change them all to
> AC_CHECK_PROG if possible. Especially the PYTHON one annoys me all the
> time. CC is another one I set occasionally.
I will object really really strongly to that, as it is 180 degrees from
where I think we need to go, and will make things a lot worse than before
on the documentation aspect that I was concerned about to begin with.
If we need to fix things so that AC_PATH_PROG will honor a non-path
input value, then let's do that. But let's not make the build system
shakier/less reproducible than it is already.
I suggest that we could inject logic like this:
if VARIABLE-is-set-and-value-isn't-already-absolute; then VARIABLE=`which $VARIABLE 2>/dev/null` fi
in front of the existing logic for AC_PATH_PROG(VARIABLE,...).
Maybe "which" isn't the best tool for the job, not sure.
Another idea, which would probably require replacing _AC_PATH_PROG
rather than just putting a wrapper around it, would be to let it
perform its normal path walk but using the given word instead of
$ac_word.
regards, tom lane