Andrew Dunstan <andrew@dunslane.net> writes:
> On 03/30/2011 12:29 PM, Dimitri Fontaine wrote:
>> Andrew Dunstan<andrew@dunslane.net> writes:
>>> I think we're pretty much down to only fixing bugs now, for 9.1, and this
>>> isn't a bug, however inconvenient it might be.
>> It's not just inconvenient, it's setting a bad example for people to
>> work on their own extensions. It's more than unfortunate. I will
>> prepare a doc patch, but copying from contrib is the usual way to go
>> creating your own extension, right?
> None of that makes it a bug.
Possibly more to the point, we don't even have a design sketch for a
better solution; and we are *certainly* past the point where blue-sky
stuff ought to be going into 9.1.
The reason it seems (to me) nontrivial to change is this: the PGXS build
method assumes that the correct pg_config can be found in your PATH.
That is pretty much guaranteed to not be the case during a in-tree
build. Even if we modified the PATH to include wherever pg_config is
hiding, the information it puts out about where to look for include and
library files would be wrong.
Another small problem with depending on pg_config during an in-tree
build is that it would completely break cross-compile builds. (Maybe
those are in bad shape already, I'm not sure. But configure for example
is still going out of its way to support them.)
So, I'm interested in trying to improve this, but it looks like a
research project from here.
regards, tom lane