From mboxrd@z Thu Jan 1 00:00:00 1970 From: dewar@gnat.com (Robert Dewar) To: Florian.Weimer@rus.uni-stuttgart.de, rms@gnu.org Cc: gcc@gcc.gnu.org Subject: Re: Why not gnat Ada in gcc? Date: Sat, 04 Nov 2000 09:04:00 -0000 Message-id: <20001104170450.BA76634D87@nile.gnat.com> X-SW-Source: 2000-11/msg00257.html <> Unfort8unately the reason we are conservative is that we have not solved the tecnhnical problems associated with the use of shared libraries for GNARL. The RPM's that are being built by others do not pass our full regression suite. Now for the informal use for which most people are using these RPM's the errors are marginal, and probably most people won't run into them, but we cannot produce RPM's that malfunction. We do not consider producing flawed software to be "supprting GNU/Linux well". We are continuing to work on the tecnbical problems here. The reason we do not default to native (kernel) threads is that there have been technnical problems of exact compliance between the GNU/Linux native thread model and the vgery specific requirements of the Ada definition. We are experimenting now with making the use of GNU/Linux native threads the default, but there is still work to be done here. Again, for the informal use to which most people put these RPM's, the differences are small, and indeed for many purposes, the native threads are preferable. Until these technical issues are resolved, we recommend that any serious use of GNU Ada use the binary versions prepared by Ada Core Technologies. Yes, we are conservative, it is definitely part of our commitment to high reliability, which after all is what Ada is all about! Robert Dewar