Martin Bonner wrote:

I think you misunderstood what Alisdair said. He offered to help REMOVE support for old borland compilers in new boost releases. The suggestion is that 1.34 supports the latest Borland compiler, but is not cluttered with workrounds for previous Borland compilers. This seems like a good compromise to me.

Precisely, although I fear that will not remove too many :( I am also in favour of '1 deprecated release' before removing support entirely, just to put users on notice. That is a separate issue though, and affects VC6/gcc2.95 just as much. There also remains the issue of test resources. If we have no-one regression testing a compiler, it de-facto loses active support so it is up to the interested community to provide support. To that end, I am trying to get the latest Borland compiler to metacomm to continue testing (with the resources they so generously provide - MUCH appreciated over the years) and working on the set of BCB2006 patches. I am reluctant to commit said patches until: i/ There are regular regression runs available for both old and new compiler ii/ I have a complete set of fixes, rather than coming back and committing a new patch-a-day. Obviously, if consensus is to drop BCB6 support in the 1.34 release, I only need a regular regression runner for the new compiler. I am looking at what I can do to fill that role myself, but ideally I would prefer to see someone elses PC testing the patches I propose, just in case there are any local config issues. I am certainly prepared to step in as a BCB2006 regression runner for the 1.34 release if that is what is needed. -- AlisdairM