Richard B. Kreckel <[EMAIL PROTECTED]> wrote: > Since the creation of the GCC 4.0 branch back in February a number of > minor C++ language changes seem to have slipped in. Let me mention just > two examples: [...]
> Are you really, really sure such language tightening is appropiate for > bug-fix releases? (Note that the examples above are not regressions since > gcc-3.4.y accept both of them.) Notice that we consider up to gcc 2.95 to check for regression status. If the code was correctly rejected in any version of GCC up to 2.95, then it is a regression, even if gcc 3.4 was wrong. But even if that couple of snippets weren't regressions, there are very high chances that a small variations of them could be a regression -- in fact, if the bug was fixed, it *must* have been a regression! Somebody found it, reported it in Bugzilla, and it got fixed. I suggest you to do some Bugzilla archeology. We don't really fix anything in dot releases unless there is a regression bug open in Bugzilla. Also, we try and keep the patch to fix the bug as minimal as possible -- sometimes, there is also a more complete and invasive patch to "properly" fix the same regression which is committed to HEAD. We can't do more than this. If a patch to fix a regression accidentally also fixes a very similar testcase which is not a regression, then let it be. -- Giovanni Bajo