3.2 PATCH: Fully support parallel gnat1/gnatbind builds

Richard Kenner kenner@vlsi1.ultra.nyu.edu
Thu May 23 14:31:00 GMT 2002


    > We rely on things besides the C ABI: for example, we found that
    > there's an incompatibility between builtin setjmp/longjmp in 2.8.1
    > and 3.2 that caused problems.

    Why is this dependency on builtin setjmp/longjmp desirable?

Because it's the simplest (and hence safest) way to implement EH.  Depending
on any *other* EH mechanism to be the same between GCC version will be much
*less* likely.



More information about the Gcc-patches mailing list