I agree that a command line option such as --config would be more in line with the Gnu way of doing things.  On the other hand spss, which we claim to emulate,  apparently puts this information under SHOW ALL.

There is no reason why we couldn't do both.

> It's true that some of the implementation detail would be exposed. Is that necessarily bad? My concern is doing it in a way which would continue to be maintainable.

I guess it wouldn't hurt anything, although a lot of users probably wouldn't understand what it meant.

Is this more useful as a runtime command or as something to show in, say, --version output?  What's the proposed use case for the information, in other words?

The COMPILE-OPTIONS that you suggest exposes a lot of implementation detail, I think.

It's not too hard to add a line like: "PostgreSQL is (not) supported," I guess.

I'm thinking about how this could best be done.
I'm a bit worried about a possible proliferation of entries in SHOW ALL.
How would it be if we had a single entry containing a concatenated string of all compile time options? 

For example

COMPILE-OPTIONS = "PSQL_SUPPORT=yes GNM_READER_SUPPORT=no ODF_FEATURES=partial" .... etc

???