chore(examples/web): use exported strings for attributes

[maryliag]

Which problem is this PR solving?

• Part Of Update usage of semantic conventions to new exported strings #2025

Short description of the changes

On package examples/web:

• Update @opentelemetry/semantic-conventions ^1.23.0
• Use exported strings for Semantic Attributes.

[codecov]

Codecov Report

Merging #2129 (eac351b) into main (dfb2dff) will decrease coverage by 0.63%.
Report is 64 commits behind head on main.
The diff coverage is n/a.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #2129      +/-   ##
==========================================
- Coverage   90.97%   90.35%   -0.63%     
==========================================
  Files         146      147       +1     
  Lines        7492     7679     +187     
  Branches     1502     1575      +73     
==========================================
+ Hits         6816     6938     +122     
- Misses        676      741      +65     

see 9 files with indirect coverage changes

[trentm]

This only requires v1.22.0, but I think we can have (perhaps even want) the examples all to be updated to be using the very latest versions of deps? Not sure how strict we need to be here. If we need to back this up to ^1.22.0 later, we can do another change.

[Flarna]

This definitely brings the example in bad state. "@opentelemetry/instrumentation-xml-http-request": "^0.39.1" depends on "@opentelemetry/semantic-conventions": "1.13.0".
So user ends up with a duplicate.

In general I think examples should be only touched if it is verified manually that result is fine as there are no automated tests.

[trentm]

Does having two versions of @opentelemetry/semantic-conventions necessarily mean a bad state for the example?

The current state already has three different versions of @opentelemetry/resource, FWIW:

[23:37:57 trentm@peach:~/src/opentelemetry-js-contrib/examples/web (git:main)]
% npm ls @opentelemetry/resources
[email protected] /Users/trentm/src/opentelemetry-js-contrib/examples/web
├─┬ @opentelemetry/[email protected]
│ ├─┬ @opentelemetry/[email protected]
│ │ └─┬ @opentelemetry/[email protected]
│ │   └─┬ @opentelemetry/[email protected]
│ │     └── @opentelemetry/[email protected]
│ └─┬ @opentelemetry/[email protected]
│   └─┬ @opentelemetry/[email protected]
│     └─┬ @opentelemetry/[email protected]
│       └── @opentelemetry/[email protected]
├─┬ @opentelemetry/[email protected]
│ ├─┬ @opentelemetry/[email protected]
│ │ ├── @opentelemetry/[email protected] deduped
│ │ ├─┬ @opentelemetry/[email protected]
│ │ │ └── @opentelemetry/[email protected] deduped
│ │ └─┬ @opentelemetry/[email protected]
│ │   └── @opentelemetry/[email protected] deduped
│ ├── @opentelemetry/[email protected]
│ └─┬ @opentelemetry/[email protected]
│   └── @opentelemetry/[email protected] deduped
└─┬ @opentelemetry/[email protected]
  └─┬ @opentelemetry/[email protected]
    └── @opentelemetry/[email protected]

and 7 installs of 3 different versions of @opentelemetry/semantic-conventions:

% npm ls @opentelemetry/semantic-conventions | rg -v deduped | rg semantic
│ │ └── @opentelemetry/[email protected]
│   └── @opentelemetry/[email protected]
│ │ └── @opentelemetry/[email protected]
│ │ └── @opentelemetry/[email protected]
│   └── @opentelemetry/[email protected]
│ └── @opentelemetry/[email protected]
└── @opentelemetry/[email protected]

Anyway, it is a bit of a mess. I'm not sure these PRs updating the examples/* make it worse, but I understand your point.

Getting into attempting to maintain many of these examples is a huge pain. Some of them are ancient or obsolete (B3Propagator). The examples/web/README.md, for example, explains nothing about what to expect, whether to run the docker-compose file that is included, etc.

[trentm]

This was discussed in the SIG today (https://docs.google.com/document/d/1tCyoQK49WVcE-x8oryZOTTToFm7sIeUhxFPm9g-qL1k/edit#heading=h.dvqq8u408ifx) and it was felt that we should go ahead with these semconv updates in the examples/* dirs. Trying to summarize the discussion from memory:

• Possible multiple separate @opentelemetry/semantic-conventions versions won't break the example apps because the semconv package has no deps.
• To the degree that users look at the examples for how to do things, it is helpful to update the semantic-conventions package usage to the new exports from v1.22.0.
• There was some expression of hope for automated verification that examples are basically working (perhaps no more than a check that they install/compile fine). Verifying beyond that might be difficult because the examples aren't generally written to be testable. Also contributor bandwidth is an issue.

[maryliag]

/easycla