diff options
| author | Kaz Kylheku <kaz@kylheku.com> | 2021-01-04 19:53:23 -0800 |
|---|---|---|
| committer | Kaz Kylheku <kaz@kylheku.com> | 2021-01-04 19:53:23 -0800 |
| commit | a0d571e13457fcad7d43b480e354df08d49a1c89 (patch) | |
| tree | 0ef20620545e8d56e82047e727d9b6bd64198ded /ffi.c | |
| parent | df9904609a72052b1014f48e4de8fa1baa74fc94 (diff) | |
| download | txr-a0d571e13457fcad7d43b480e354df08d49a1c89.tar.gz txr-a0d571e13457fcad7d43b480e354df08d49a1c89.tar.bz2 txr-a0d571e13457fcad7d43b480e354df08d49a1c89.zip | |
configure: tweak test for inline.
It seems that the Brew people are running into a problem. The
configure test is detecting that "inline" can be used for
declaring inline functions, but then some functions are not
inlined, requiring an external definition.
On GNU/Linuxes, we detect "static inline", so this doesn't
reproduce.
This is likely triggered by the recent change to compile as
C99 (unless configuring in maintainer mode).
* configure: Let's revise the set of possible definitions of
INLINE that we try. Firstly, we only try "inline" if we are
compiling as C++. If that fails, all the other forms of inline
we try include "static". Let's not mess with anything involving
"extern". Like a bare "inline", it might work in the test
program but cause some sort of failure.
Diffstat (limited to 'ffi.c')
0 files changed, 0 insertions, 0 deletions