On 05/16/2017 07:44 PM, Peter Eisentraut wrote:
> On 5/3/17 15:14, Andrew Dunstan wrote:
>> Can someone please explain to me why we have this in Makefile.global.in?
>> (from commit e9c81b60 )
>>
>> PROVE_FLAGS =
>>
>> ISTM it's unnecessary, and prevents us from using the same named value
>> in the environment. I want to be able to use the environment in
>> vcregress.pl, and I'd like the Make files to work the same way.
> I think relying on environment variables like this is a bad design.
> Someone could have something left in the environment and it changes
> things in mysterious ways. For all other *FLAGS variables, we
> explicitly set them in Makefile.global, sometimes to empty values, as
> the case may be.
>
> Note that specifying a variable on the make command line overrides
> settings in the makefile under certain circumstances, which is a useful
> feature.
>
> So I think the above setting was correct, the new behavior is
> inconsistent and incorrect, and I think this change should be reverted.
>
It would have been nice if you'd spoken up when the topic was raised.
This doesn't rely on the environment variable, it just enables it. You
can still say "make PROVE_FLAGS=--whatever" and it will override the
environment.
Inheriting variables from the environment is a part of make by design.
We have PG_PROVE_FLAGS for our own forced settings.
Note that the previous setting was done without any thought being given
to how this works with vcregress.pl. I have been trying to make the two
systems more consistent. This was a part of that effort.
cheers
andrew
--
Andrew Dunstan https://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
