
Dear Boost Developers, I realize that I cannot ask you to support old compilers, but I would like to kindly ask you not to intentionally rule them out and to retain existing work-arounds for older compilers, if not otherwise than by separate folders and files containing "last version known to work". I think that newer versions of the libraries should *improve* support for all compilers instead of dumping a lot of work that has been put, by the original developers and by others, into getting them to work. And many of them *do* work, because for example even if they do not support Template Partial Specialization in *all* cases, they do support it in enough cases. And who supports, even that, in *all* cases ? At some time I had Boost.Thread working just fine with Borland, and the changes were two or three lines, but there was no answer from the developer to incorporate them for weeks... I gave up. Some seem to recommend dropping support for Borland C++ Builder (version 5, but as I am going to show you really soon, there is no much difference between version 5.5 aka 2000 and version XE5 aka 2013). But on older hardware (some of us do not have a job and cannot afford i7's), Borland (any version since 2000) is 2-or-3-or-4 times faster than Visual C++ (any version since 2003). As the list of actively developed native C++ compilers gets shorter and shorter (msvc, gcc, clang, dmc and just who else for Windows if I may ask ?), having more compilers to: - test our code on and - help us find bugs and - help us achieve portability and - help us achieve independence from one (commercial or otherwise) vendor are great things, very highly appreciated by many developers. Yes, my answer may be late. But I have humbly said it before: Maybe it is not the compilers failing Boost. Maybe it is Boost failing the compilers. Happy New Year ! -- Yours truly, Adder On 10/15/13, Peter Dimov <lists@pdimov.com> wrote:
Robert Ramey wrote:
I'm willing to concede that these points have value.
But I don't think the cost has been adequately appreciated.
Exactly. The cost will only become to be appreciated when a release containing the changes goes out. The majority of our users do not use the trunk, do not try out a beta release, do not try the release candidates; many of them nowadays don't even get Boost from us. There are also people who are using (portions of) Boost on "non-supported" compilers. We only hear
from them when something breaks.
_______________________________________________ Unsubscribe & other changes: http://lists.boost.org/mailman/listinfo.cgi/boost