Vladimir Prus wrote:
Thomas Witt wrote:
Hmm, then what is the point of testing RC3? It does not have neither additional bugs, nor additional fixes relative to RC2.
There is more to a release than just code. I might have screwed up packaging/html generation version numbers and so forth.
That is your pick. But I'd like to point out two things:
1. Both issues were reported in reply to your RC2 announcement. I did not saw any comment from you on those issues until now, after RC3. If bug reports for release candidates are silently ignored, what's the point of release candidates? Is there any point to report possible bugs in RC3 at all, if "time for code changes has passed"?
See my comment above.
2. Those issues are not so minor.
At som point they have to be catastrophic to justify changes. They are not.
Note that quite some effort was spend on improving cygwin support. In fact, I believe Dave spent a couple of weeks on Boost.Python cygwin support. After that effort, releasing 1.34.1 with cygwin not working makes little sense. Note that boost.filesystem issue means you cannot use convenience.hpp in more than one module; essentially part of library is not usable in real-world programs, on all compilers.
I know, there was a time to spot and fix it. It wasn't fixed. It's sad but as things are we'll never ship if we try to be perfect. Thomas -- Thomas Witt witt@acm.org