On Oct 5, 2006, at 9:21 AM, Arkadiy Vertleyb wrote:
"Michael Goldshteyn" <mgoldshteyn@comcast.net> wrote
434 failures in 15 libraries filesystem (5) graph (1) io (1) iostreams (14) mpl (4) parameter (11) program_options (9) python (214) rational (5) serialization (16) spirit (5) test (2) tr1 (56) utility (1) xpressive (90)
Well, it's pretty much a given that python, tr1, and xpressive are holding back 1.34. So, what is the prognosis on each of these? Is this build ever going to happen, or are we headed for 2007?
Well, if you take a look at the *latest* regression results, it's much better. In fact, I think xpressive should be failure-free once the last of tuesday tests is re-run, and python has only a few failures.
How does this work? Are we going to wait for zero failures? Often fixing two failures may be more difficult than fixing 400...
Usually we wait until we hit zero failures. That doesn't mean that everything works on every compiler; often, it means that we've explicitly marked up some failures as "expected," because they are too hard to fix on a given compiler. Some of the sun-5.8 ICEs would fall into that category, I expect. Cheers, Doug