Обсуждение: ecpg_config.h symbol missing with meson
I checked the generated ecpg_config.h with make and meson, and the meson one is missing #define HAVE_LONG_LONG_INT 1 This is obviously quite uninteresting, since that is required by C99. But it would be more satisfactory if we didn't have discrepancies like that. Note that we also kept ENABLE_THREAD_SAFETY in ecpg_config.h for compatibility. Fixing this on the meson side would be like diff --git a/src/interfaces/ecpg/include/meson.build b/src/interfaces/ecpg/include/meson.build index 31610fef589..b85486acbea 100644 --- a/src/interfaces/ecpg/include/meson.build +++ b/src/interfaces/ecpg/include/meson.build @@ -12,6 +12,7 @@ ecpg_conf_keys = [ ecpg_conf_data = configuration_data() ecpg_conf_data.set('ENABLE_THREAD_SAFETY', 1) +ecpg_conf_data.set('HAVE_LONG_LONG_INT', 1) foreach key : ecpg_conf_keys if cdata.has(key) Alternatively, we could remove the symbol from the make side.
Peter Eisentraut <peter@eisentraut.org> writes: > I checked the generated ecpg_config.h with make and meson, and the meson > one is missing > #define HAVE_LONG_LONG_INT 1 > This is obviously quite uninteresting, since that is required by C99. > But it would be more satisfactory if we didn't have discrepancies like > that. Note that we also kept ENABLE_THREAD_SAFETY in ecpg_config.h for > compatibility. > ... > Alternatively, we could remove the symbol from the make side. Think I'd vote for removing it, since we use it nowhere. The ENABLE_THREAD_SAFETY precedent feels a little bit different, since there's not the C99-requires-the-feature angle. regards, tom lane
On 17.04.24 18:15, Tom Lane wrote: > Peter Eisentraut <peter@eisentraut.org> writes: >> I checked the generated ecpg_config.h with make and meson, and the meson >> one is missing > >> #define HAVE_LONG_LONG_INT 1 > >> This is obviously quite uninteresting, since that is required by C99. >> But it would be more satisfactory if we didn't have discrepancies like >> that. Note that we also kept ENABLE_THREAD_SAFETY in ecpg_config.h for >> compatibility. >> ... >> Alternatively, we could remove the symbol from the make side. > > Think I'd vote for removing it, since we use it nowhere. > The ENABLE_THREAD_SAFETY precedent feels a little bit different, > since there's not the C99-requires-the-feature angle. Ok, fixed by removing instead.