STYLE: Remove std:: prefix from types that work without it

[Leengit]

This is accomplished with

cd git/ITK
find * -type f |
  egrep '\.(c|cc|cxx|cpp|h|hh|hxx|hpp)$' |
  fgrep -v ThirdParty |
  fgrep -v itkIntTypes.h |
  xargs sed -i -r -e \
    's/std::((size|ptrdiff|max_align|u?int(_fast|_least|max|ptr)?[0-9]*)_t)/\1/g'

Closes #3248.

PR Checklist

• No API changes were made (or the changes have been approved)
• No major design changes were made (or the changes have been approved)
• Added test (or behavior not changed)
• Updated API documentation (or API not changed)
• Added license to new files (if any)
• Added Python wrapping to new files (if any) as described in ITK Software Guide Section 9.5
• Added ITK examples for all new major features (if any)

[N-Dekker]

@Leengit C++ doesn't specify whether or not the std integer types from <cstdint> are also defined in the global namespace. Is the idea for ITK code to always use the aliases from the itk namespace, defined by "itkIntTypes.h"?

[Leengit]

Is the idea for ITK code to always use the aliases from the itk namespace, defined by "itkIntTypes.h"?

Wherever they come from doesn't matter to me. Folks have been using the short form predominantly, it works, and I am bringing the minority of cases into the fold.

[N-Dekker]

Wherever they come from doesn't matter to me. Folks have been using the short form predominantly, it works, and I am bringing the minority of cases into the fold.

The majority isn't always right! It does help if we could specify why and when the std:: may be dropped. Because we won't remove all std:: in the entire code base, right?

There is no "max_align..." alias in "itkIntTypes.h", should it still be added?

[dzenanz]

integer types from <cstdint> are also defined in the global namespace

I understood this is true from https://en.cppreference.com/w/cpp/types/integer.

[Leengit]

Good point about max_align_t, @N-Dekker. I'd say defer. This pull request conversation is hereby amended with the following disclaimer.

Some of these types are in the global namespace and some are in the itk namespace due to itkIntTypes.h, or are in both. Some of the types listed in the sed code snippet above might be in neither, but they are not causing problems because they are not currently used in our code base. If the short form for one of these is used later and it fails then it may make sense to add it to itkIntTypes.h.

[Leengit]

The majority isn't always right!

What do you propose?

[N-Dekker]

The majority isn't always right!
What do you propose?

In general: if there is a significant minority of the code doing it otherwise, double-check if the majority of the code really does the right thing, before adjusting the minority to the majority.

Specifically in this case, it appears reasonable to remove std:: within the ITK code base, for those types that have an alias in "itkIntTypes.h". 👍

[N-Dekker]

@dzenanz

integer types from <cstdint> are also defined in the global namespace

I understood this is true from https://en.cppreference.com/w/cpp/types/integer.

The latest draft of the C++ Standard section [headers] says:

Except as noted in Clause 16 through Clause 32 and Annex D, the contents of each header cname is the
same as that of the corresponding header name.h as specified in the C standard library (Clause 2). In the
C++ standard library, however, the declarations (except for names which are defined as macros in C) are
within namespace scope (6.4.5) of the namespace std. It is unspecified whether these names (including any
overloads added in Clause 17 through Clause 32 and Annex D) are first declared within the global namespace
scope and are then injected into namespace std by explicit using-declarations (9.9).

HTH, Niels