This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: 3.3-branch QA assessment
- From: Wolfgang Bangerth <bangerth at ticam dot utexas dot edu>
- To: jbuck at synopsys dot com, <gcc at gcc dot gnu dot org>
- Date: Mon, 27 Jan 2003 18:03:29 -0600 (CST)
- Subject: Re: 3.3-branch QA assessment
Joe,
as you already note, regressions don't show the whole picture. This is to
say that at present the C++ frontend would look much better even than 3.3:
the new parser has allowed to close about 200 reports; on the other hand,
many new regressions are found, and being marked as regressions. Thus,
many old non-regressions have been converted in many (but fewer)
regressions. If these regressions continue to be fixed at the present
astonishing rate, 3.4 will be much better in terms of C++ than any release
before, although the number of regressions doesn't show this.
On the other hand, the number of regressions being found in the new parser
is large, and this will certainly not change in the near future. At least
in the case of my software, front-end bugs also still prevent me from
exercising the back-end, so I can't check the negative impacts of all the
other merges. 3.4 definitely needs some maturing.
Whether in the light of the upcoming C++ improvements anyone would
appreciate a 3.3 is certainly a good question. I for one would rather have
developers spend their time fixing bugs in mainline, than duplicating
efforts. But then, experience shows that only an upcoming release really
leads to much bug-fixing activity, so a 3.3 release also serves 3.4.
W.
-------------------------------------------------------------------------
Wolfgang Bangerth email: bangerth@ticam.utexas.edu
www: http://www.ticam.utexas.edu/~bangerth/