|
Boost : |
From: Edward Diener (eldiener_at_[hidden])
Date: 2021-01-06 21:13:47
On 1/6/2021 3:10 PM, René Ferdinand Rivera Morell via Boost wrote:
> On Wed, Jan 6, 2021 at 2:02 PM Edward Diener via Boost <
> boost_at_[hidden]> wrote:
>
>>
>> If an end-user does not see a "C++ standard minimum level" for a library
>> in the documentation what should he/she assume ?
>
>
> They should *not* assume any particular level. And perhaps ask the author
> to indicate what the support is by filling GitHub issues.
The trouble with this is that there are about 110 libraries with no
'cxxstd' JSON attribute at all and we are asking programmers to go back
to guessing what the C++ standard minimum level is for those libraries.
I am really trying to give end-users information rather than going back
to guessing or having to ask about it each time.
>
>
>> I would like to argue
>> that it would be beneficial for end-users to see the C++ standard
>> minimum level for all libraries, even for those which work at the
>> C++98/03 level on up. I do not understand the purpose of a library which
>> does not wish to peg itself at a particular minimum C++ level.
>>
>
> At least one library supports *any* C++ level. Even those before C++98.
> Please don't put a one dimensional straightjacket on this. What should such
> libraries do?
>
> PS. That one library happens to be my library, Predef.
You added "cxxstd": "98" for predef. What is wrong with that ? It
basically says "any" C++ level. I would also argue that "03" essentially
say "any" C++ level, because how many programmers are using C++ before
it was standardized and how many programmers can distinguish between
C++98 and C++03 ? But I have no beef with "98" since that is what is
there. You are arguing that showing nothing is somehow better than
showing something, and I admit I do not see that.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk