[boost] A possible date for dropping c++03 support

Subject: [boost] A possible date for dropping c++03 support
From: Mike Dev (mike.dev_at_[hidden])
Date: 2018-08-28 05:11:15

Dear Edward,

throughout this thread you keep saying that you or the user will 
not understand what 

   "dropping c++03 support" 

means and I have to say I find that very hard to believe, as the 
concept of "XXX is not supported" is ubiquitous throughout software 
development. 

There are of course differences in detail, but 
universally it always means something along the lines of

"We don't promise that XXX works. 
 When you try XXX you are on your own, we take no responsibility 
 for what happens if you do and even if it works now, it might 
 not work with the next product version (even if it is just a 
 minor/path update)."

Now, on top of that, "XXX is not supported" often means that 
"XXX really just doesn't work", but I can give you dozens 
(probably thousands) of examples where things that are not 
officially supported happen to work e.g.:

- Forward declaring anything from namespace std or defining 
  anything inside std is UB, yet many libraries (inside and 
  outside of boost) do it anyway and "it works"

- Microsoft's WSL currently doesn't support running graphical 
  applications, however, users have been using it to successfully 
  run various Linux DMs for quite some time now

- Many libraries (inside and outside of boost) have a list 
  of supported compilers. Does that mean they don't work with 
  any compiler not mentioned on that list? Probably not (if I 
  try a newer version of gcc or msvc, it most of the time still 
  works) - they are just the compilers the author cares about 
  and/or has tested so far.

- Abseil.io has a very clear policy of what platforms and compilers 
  are supported (https://abseil.io/docs/cpp/platforms/platforms) 
  but again, there exist other platform/compiler combinations that 
  will probably work just as well. 

- Intel's drivers for consumer lan chips don't support windows 
  server, but if you manually circumvent the check, they work 
  quite well.

So really, the concept of it happens to work but isn't supported 
shouldn't be foreign to anyone active in software development or
IT in general.