[SEMANTIC CONVENTIONS] Upgrade to version 1.15.0

[marcalff]

Fixes #1758

Changes

Upgrade semantic conventions to opentelemetry-specification 1.15.0
Change naming to use camel case, to avoid collisions with macros.

• CHANGELOG.md updated for non-trivial changes
• Unit tests have been added
• Changes in public API reviewed

[codecov]

Codecov Report

Merging #1761 (e1945b2) into main (a3e3a90) will not change coverage.
The diff coverage is 100.00%.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1761   +/-   ##
=======================================
  Coverage   85.77%   85.77%           
=======================================
  Files         171      171           
  Lines        5214     5214           
=======================================
  Hits         4472     4472           
  Misses        742      742           

Impacted Files	Coverage Δ
sdk/src/resource/resource.cc	96.30% <100.00%> (ø)

[lalitb]

Thanks for the upgrade.

[marcalff]

@lalitb

Please note: I had to implement a work around for ERROR in windows, similar to DELETE:

#ifdef _WIN32
#  ifdef ERROR
#    pragma message(__FILE__ ": removing define on ERROR")
#    undef ERROR
#  endif
#endif

Build on windows fails without it.

[ThomsonTan]

@lalitb

Please note: I had to implement a work around for ERROR in windows, similar to DELETE:

#ifdef _WIN32
#  ifdef ERROR
#    pragma message(__FILE__ ": removing define on ERROR")
#    undef ERROR
#  endif
#endif

Build on windows fails without it.

Wondering why this is needed now.

[lalitb]

@lalitb
Please note: I had to implement a work around for ERROR in windows, similar to DELETE:

#ifdef _WIN32
#  ifdef ERROR
#    pragma message(__FILE__ ": removing define on ERROR")
#    undef ERROR
#  endif
#endif

Build on windows fails without it.

Wondering why this is needed now.

I think this line is having conflict with ERROR macro

static constexpr const char *ERROR = "ERROR";

[ThomsonTan]

@lalitb
Please note: I had to implement a work around for ERROR in windows, similar to DELETE:

#ifdef _WIN32
#  ifdef ERROR
#    pragma message(__FILE__ ": removing define on ERROR")
#    undef ERROR
#  endif
#endif

Build on windows fails without it.

Wondering why this is needed now.

I think this line is having conflict with ERROR macro

static constexpr const char *ERROR = "ERROR";

Can we rename the new variable here?

[lalitb]

Can we rename the new variable here?

Yeah good point, if it is possible to do in the jinja template.

[marcalff]

Some comments about ERROR:

• I don't know what the ERROR macro is on Windows, and what it is for. So, no idea if undef ERROR is intrusive or not for the application code.
• It is confirmed, from the build logs, that there is an ERROR macro somewhere, see bazel windows.
• Could someone point out which header file on windows is defining an ERROR macro, will be needed in the comments.

If we rename it:

• Any proposal about the name ?
• Should we have some consistency about alternate names, like SEMCONV_<foo> to replace <foo> on name collisions ? It is likely that more collisions will happen again, with new semantic conventions.
• Should the alternate name, say SEMCONV_ERROR, be used on the offending platform only or on all platforms ?
• Should we change undef DELETE to SEMCONV_DELETE then ?

My preference at this point is:

• change DELETE to SEMCONV_DELETE
• change ERROR to SEMCONV_ERROR
• apply for all platform, so user code that builds on multiple platforms uses the same symbol in all cases.
• never undef anything, to avoid being intrusive for applications, even when they do not use the offending semantic convention (this is a change compared to previous solution for DELETE)

Requesting re-review to settle this.

[lalitb]

@marcalff Should we instead change the variable name to follow the C++ style guide (https://google.github.io/styleguide/cppguide.html#Constant_Names) i.e, k<camelCase>. I see the to_camel_case method defined in templating code, not sure if we can use it and prefix k before it.

Also semantic conventions specs are still experimental, so we can use that to our benefit :)

[marcalff]

@marcalff Should we instead change the variable name to follow the C++ style guide (https://google.github.io/styleguide/cppguide.html#Constant_Names) i.e, k<camelCase>. I see the to_camel_case method defined in templating code, not sure if we can use it and prefix k before it.

Also semantic conventions specs are still experimental, so we can use that to our benefit :)

Thanks @lalitb, this is a much better solution.

Implemented.

[lalitb]

LGTM. Thanks :)