Re: s_lock.h default definitions are rather confused

Andres Freund <andres(at)2ndquadrant(dot)com> writes:
> Given you got the error above, you used gcc. Have you used non-gcc
> compiler on hppa recently? I seem to recall you mentioning that that
> doesn't work sanely anymore? If so, perhaps we can just remove the !gcc
> variant?

It still compiles, modulo some old and uninteresting warnings, but linking
the postgres executable fails with

usr/ccs/bin/ld: Unsatisfied symbols:
pg_compiler_barrier_impl (code)
make[2]: *** [postgres] Error 1

Curiously, there are no compiler warnings about "reference to undeclared
function", which is odd because I see nothing that would declare that name
as a function or macro for non-gcc HPPA. But in any case, the fallback
logic for compiler barriers evidently still needs work.

... further on ... Looks like somebody has recently broken pg_dump, too:

cc: "pg_dump.c", line 319: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 320: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 321: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 322: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 323: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 324: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 326: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 327: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 329: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 333: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 335: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 336: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 337: error 1521: Incorrect initialization.
cc: "pg_dump.c", line 338: error 1521: Incorrect initialization.
make[3]: *** [pg_dump.o] Error 1

This one looks like overoptimistic assumptions about what's legal
in an initializer. We could probably fix that by reverting the
option variables to statics; I see no benefit to be had from
having them as dynamic variables in main().

regards, tom lane