On Sun, May 3, 2020 at 5:25 AM Bruno Haible <[email protected]> wrote: > > Jeffrey Walton wrote: > > Maybe configure should die if LDBL_MANT_DIG is not expected. That > > should avoid the problem of silent failures. When configure fails > > folks will have to tend to the problem. > > I disagree. We have unit tests. For many packages, math.h related test > failures - and especially 'long double' related test failures - are > acceptable. The person who runs "make check" can surely evaluate the > severity of a test failure. > > Having configure die is the worst possible behaviour, because it elevates > the issue to severity 1 / BLOCKER.
The logic assumes everyone runs 'make check'. I know for certain it is not the case. That leaves the silent failures as the default use case for some users. (I don't claim users who fail to run 'make check' are correct. I just acknowledge they exist and you should engineer around them). Jeff
