pkgsrc-Changes archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: CVS commit: pkgsrc/cad



nia <nia%NetBSD.org@localhost> writes:

> Hey, so for my recent commits what I did was:
>
> - convert GCC_REQD to USE_CXX_FEATURES
> - if I could, attempt to build it with default compiler settings
>   (i.e. USE_LANGUAGES set to c++03) and see if it built. If it does,
>   also remove the version from USE_LANGUAGES if USE_CXX_FEATURES
>   exists.

So in every case, if c++17 is in USE_LANGUAGES, then c++17 ends up in
USE_CXX_FEATURES?   So

  USE_LANGUAGES+=c++17
  GCC_REQD+=5 #or whatever

leads to

  USE_LANGUAGES+=c++
  USE_CXX_FEATURES+=c++17

and

  USE_LANGUAGES+=c++17

leads to

  USE_LANGUAGES+=c++
  USE_CXX_FEATURES+=c++17

both?

What I think is not ok is:

  USE_LANGUAGES+=c++17

leading to

  USE_LANGUAGES+=c++

without adding to USE_CXX_FEATURES, because that drops the information
that a c++17 compiler is required -- even if it never forced it -- and
it has the unintended effect of not failing early due to --std being
inserted, and instead relying on the compiler throwing errors from
constructs that are not c++03.

On rereading the current state of kicad, I see c++17 is still in
USE_LANGUAGES, so that's merely an incomplete conversion, on the basis
of caution that it might require std forcing.

Is that the way you see it?


> The behaviour of USE_LANGUAGES is unchanged in pkgsrc - from
> your email Greg, it sounds like I changed what it does. I changed
> the way it's documented only.

Sorry, I didn't mean to imply that.  What I thought was happening was
c++17 being dropped from various packages' USE_LANGUAGES without being
added to USE_CXX_FEATURES (absent a determination that the prior usage
was truly buggy).



Home | Main Index | Thread Index | Old Index