http://gcc.gnu.org/bugzilla/show_bug.cgi?id=51616
Tobias Burnus <burnus at gcc dot gnu.org> changed: What |Removed |Added ---------------------------------------------------------------------------- CC| |burnus at gcc dot gnu.org --- Comment #9 from Tobias Burnus <burnus at gcc dot gnu.org> 2012-01-12 07:57:53 UTC --- Please confirm that it no longer fails. There is now a dg-effective-target check which does not run the test case if for the highest available floating-point data type (here: REAL(16)) no sqrt function is available. Thus, the test case will no longer fail as it is not run. (In reply to comment #7) > There was some discussion last year about adding support > to provide `l' aliases to the `q' routines in libquadmath. I think one could add - for HPUX - a configure test whether (a) libquadmath can be build, (b) whether "long double" and __float128 are the same, and (c) whether "sqrtl" is not available. If this is fulfilled, one could provide the symbols in libquadmath. But that's a different issue and should be started by someone who has access to such a system.