Re: Building some extra libs (renaissance, gnustep-guile,

[Yavor Doganov]

В Fri, 02 Oct 2009 19:39:44 +0100, Richard Frith-Macdonald написа:

> Based on a lot of trial and error, it seems that --disable-jdbc-bundle
> triggers some autoconf bug such that subsequent tests fail.

Not an Autoconf bug.  With the goal to provide smaller configure scripts 
and avoid duplicate checks, code for AC_REQUIRE'd macros is not inserted 
on every invocation of a particular macro.  IOW, AC_CHECK_HEADERS 
indirectly requires AC_PROG_CC and AC_PROG_CPP, but code for checking the 
compiler/preprocessor is not inserted at every place you call that macro, 
only before the first expansion of AC_CHECK_HEADERS (or any other macro 
that requires these variables to be set).  Autom4te discovers if/when it 
is needed by tracing configure.ac.

Because with --disable-jdbc-bundle the macro expansion falls in the 
`else' branch, the code for detecting the compiler/preprocessor is not 
executed and every further test that relies on these variables being set 
is broken.

> I have put a workaround in configure.ac in svn trunk (adding an extra  
> call to AC_CHECK_HEADERS before the test

In case of nested if/else statements like here, and in general almost 
always, the proper fix is to call AC_PROG_CC and AC_PROG_CPP earlier in 
configure.ac, before any conditionals.