zpp_bits: Add new wrap 4.4.22

[QWERTIOX]

I'm not sure if meson_version: '>=1.0.1' should be so high but this is the one i'm using right now

[QWERTIOX]

of course sanity checks pass on local machine...

[benoit-pierre]

You forgot to update releases.json.

As for the meson_version: '>=1.0.1' requirement, there's no reason for it.

[benoit-pierre]

This file must be removed.

[QWERTIOX]

wait, i include whole directory and not just .h file

[benoit-pierre]

Why?

[QWERTIOX]

because it was including junk like readme.md and I could't find another way to move this header file or force including just header file

[benoit-pierre]

If by included, you mean add the directory to the include flags, yes… So? You're afraid someone is going to use #include <README.md>?

And you could have just used fs.copyfile to copy the include to an include sub-directory at setup.

[QWERTIOX]

i tried and it didn't see the include directory

[benoit-pierre]

Please post the code you used.

[QWERTIOX]

project(
  'zpp_bits',
  'cpp',
  version: '4.4.22',
  license: 'MIT',
)
fs = import('fs')
copy = fs.copyfile('zpp_bits.h', 'include')

incdir = include_directories('include')

zpp_bits_dep = declare_dependency(include_directories: include_directories())

[benoit-pierre]

That's not how fs.copyfile work:

dest str: the name of the output file. If unset will be the basename of the src argument

Try something like this:

• meson.build

project(
  'zpp_bits',
  'cpp',
  version: '4.4.22',
  license: 'MIT',
  meson_version: '>=0.64.0',
)

subdir('include')

zpp_bits_dep = declare_dependency(include_directories: incdir)

• include/meson.build:

copy = import('fs').copyfile('../zpp_bits.h')

incdir = include_directories('.')

[benoit-pierre]

(And go back to the original upstream source distribution).

[QWERTIOX]

i think meson.build will still be included this way

[benoit-pierre]

What do you mean included? declare_dependency is not creating an archive.

[QWERTIOX]

i mean it's in the same directory as header

[benoit-pierre]

IHMO, you could have kept the original code. The real use case for fs.copyfile with respect to headers is to avoid making private headers accessible when they are mixed with public ones, particularly things like headers with generic names, like config.h.

i mean it's in the same directory as header

So? You'll have multiple instances of that in the wraps provided (with makefiles, source files…). Unless someone is foolish enough to try to include<meson.build>, what is the harm exactly?

[QWERTIOX]

maybe i'm perfectionist but i don't like that my tools reports that meson.build is one of the headers ready to be included

[QWERTIOX]

but as you mentioned, other wraps do that so now i don't care

[QWERTIOX]

of course i'm stupid - i forget to set c++ at least to c++20

[neheb]

uhhh

[QWERTIOX]

I don't know why it did that, maybe because it's second to last so maybe it's some problem with trailing comma that makes diff freak out

[benoit-pierre]

No, you removed the terminating newline.

[QWERTIOX]

Was that bad?