Peter Dimov wrote:
Am I the only one disturbed by the difficulties we have in getting an "official stable release" out? :-)
See below where I am taking current RC from CVS to migrate to TR1. [There are timing reasons why migrating this week is very important for us - hopefully updating to the final release will be comparatively painless] The main issue seems to be with new tests/platforms failing, rather than regressions. I don't see much sign of progress/patching, so are we just waiting for the failures to be marked up? I guess more 'platform experts' need to bite the bullet and admit that their compilers are not going to handle all the new tests for this release. If we start marking known failures where no-one is working on patches, I expect the release could look much closer by the end of the week. On the Borland front - I want to investigate the numeric_cast failures with BCB2006, but suspect most other failures can be marked as expected fails now. I had just been a little slow hoping we might find another workaround or two while the other platforms stabalised. Guess everyone else is thinking the same? [I have stayed away from Parameter library which seems to have a lot of new fails on 'legacy' platforms. If Dave has no plans to investigate these I would suggest marking all Metacomm fails on this library as 'expected'] -- AlisdairM