On Wed, Jul 9, 2025 at 12:07 PM Tom Lane <t...@sss.pgh.pa.us> wrote: > Jacob Champion <jacob.champ...@enterprisedb.com> writes: > > Here is a draft patch for Ivan's reported issue; I still need to put > > it through its paces with some more unusual setups, but I want to get > > cfbot on it. > > I'm confused about why this moves up the temporary changes of > CPPFLAGS and LDFLAGS, but not LIBS? Maybe that's actually correct, > but it looks strange (and perhaps deserves a comment about why).
Yeah, that's fair. It's because LIBCURL_LDLIBS isn't set until that AC_CHECK_LIB test is run, and the test needs LIBCURL_LDFLAGS to be in force. (Upthread, I was idly wondering if those AC_CHECKs should just be removed -- after all, PKG_CHECK_MODULES just told us where Curl was -- but I'm nervous that this might make more niche use cases like cross-compilation harder to use in practice?) Does the attached help clarify? Thanks, --Jacob
v2-0001-WIP-oauth-run-Autoconf-tests-with-correct-compile.patch
Description: Binary data
