Marsh Ray wrote:
On 10/04/2010 03:20 AM, John Maddock wrote: The idea is that if a Boost developer happens to be working in a source file and comes across a block of VC6 workaround code that he'd really like to whack, he can just go ahead and whack it knowing he has the full support of the users behind him.
That's exactly what's happening at least in my case. MSVC 6.0 isn't being tested anymore. So any change I make is likely to break it. I seriously doubt that the current serialization library would compile with this compiler. With Borland, I don't have a borland compiler, but even though it's being tested, I couldn't maintain compatibility and anyway it's not worth it. If someone wanted to suggest changes (and run tests) to maintain support for these compilers, I would be willing to roll them in - but I'm not spending my own time on this. Basically, I respond to "market" demand. The problem with this is that since the "product" isn't sold I have no way to gauge this demand other than these lists. This is a large part of the motivation behind the proposal I presented at BoostCon this year. Robert Ramey PS. I think I may have misunderstand the original proposal. If its to strip out compatibility code for these compilers my vote would be -1 If it would be to just stop investing effort in maintaining compatibility - which is the current policy - I would vote +1 RR