Re: FORCE_C_STD and FORCE_CXX_STD

[Greg Troxel <gdt%lexort.com@localhost>]

Edgar Fuß <ef%math.uni-bonn.de@localhost> writes:

> The proposal seems to be lacking the distinction between "I need at least N" 
> and "I don't work with newer than N". This appears (to me) to be relevant 
> in case multiple such requirements exist (due to buildlink3, for example). 
> Or isn't that supposed to happen?

USE_CXX_FEATURES is for requiring a set of language variants.

This is for adding --std=foo, which is IMHO by definition working around
a bug in an upstream program, which should probe for and use the --std
for the language variant it needs.

This is a patch appearing for the first time for review, but we have
extensively discussed the concept, which is moving --std forcing form
the (misnamed IMHO) USE_LANGUAGES into a new variable.   This is a
similar in spirit cleanup to defining USE_CXX_FEATURES.

My understanding is that proposed change is not going to change the way
any builds actually happen, but it will significantly clean up how
"force std" is expressed and make things much easier to understand.
That should increase the odds of such things being expressed correctly;
the situation before USE_CXX_FEATURES was definitely too confusing (and
didn't allow expressing what needed to be expressed in a way that worked).