[chore] Update vertex_ai to vertex.ai

[gyliu513]

Fixes #

This is a follow up for #1574 (comment)

/cc @lmolkova

Changes

Please provide a brief description of the changes here.

Note: if the PR is touching an area that is not listed in the existing areas, or the area does not have sufficient domain experts coverage, the PR might be tagged as experts needed and move slowly until experts are identified.

Merge requirement checklist

• CONTRIBUTING.md guidelines followed.
• Change log entry added, according to the guidelines in When to add a changelog entry.
  • If your PR does not need a change log, start the PR title with [chore]
• schema-next.yaml updated with changes to existing conventions.

[lmolkova]

Thanks @gyliu513 !

Please make sure to add changelog entry.

/cc @aabmass - wdyt about this rename?

[gyliu513]

@lmolkova Done!

But seems the CI is reporting an compatibility issue:

Found 1 compatibility issues:
	enum attribute member 'gen_ai.system.vertex_ai' was removed
make: *** [Makefile:276: compatibility-check] Error 1

[gyliu513]

@lmolkova do you know if we can also need to mark vertex_ai as deprecated? if so, how to make it? This is a value and not an attribute, did not find any example for this, thanks!

[gyliu513]

I will update https://github.com/open-telemetry/semantic-conventions/blob/main/schema-next.yaml after #1683 got merged.

[aabmass]

I can't find any evidence of just "Vertex" without "AI" in GCP's docs, so vertex_ai actually seems appropriate to me. I can't imagine any other vertex.* values.

If that reasoning holds water to keep it as vertex_ai, I can verify with the team.

[joaopgrassi]

I also believe we should not change this to be a namespace. Vertex AI seems to be the product name.

[joaopgrassi]

OK I re-read and thought about it, giving this is only changing the enum value all if fine.

[aabmass]

While we're here, should we prefix with gcp. like we have for some other vendors?

[gyliu513]

+1, we can name it as gcp.vertex.ai or google.vertex.ai

[lmolkova]

we're really inconsistent in semconv on prefixing things with cloud name (not intentionally). General issue - #608

E.g. we have spanner and it's ok since there is just one spanner DB.
We have gcp_pubsub for messaging because pubsub is too generic. We also have some gcp.* attributes and resources.

Both vertex.ai or vertext_ai are unique enough - they don't have to have a prefix similarly to spanner. If we add prefix, it should be gcp.

So what should we do:

1. spanner, vertex_ai, gcp_pubsub
2. spanner, vertex_ai, gcp.pubsub
3. gcp.spanner, gcp.vertex_ai, gcp.pubsub
4. gcp_spanner, gcp_vertex_ai, gcp_pubsub
5. ...
   ?

Speaking from Azure side, we're likely to stay where we are: cosmosdb, eventhubs, az.ai.inference (use product name when it's unique enough, use az. namespace when it's not). Reasons: it's nice to be consistent, but some of these things are de-facto stable and are used wide enough to make breaking really problematic.

cc @jsuereth if he has any thoughts

[aabmass]

There's also the cloud.platform values which look like azure_functions, gcp_cloud_functions .

[lmolkova]

I'm like 80% confident we can rename azure_functions to az.functions. Let me try

[joaopgrassi]

I'd leave the final to the experts and product folks, but to my understanding, Vertex AI is a trademark https://trademarks.justia.com/907/18/vertex-90718213.html, so making it vertex.ai is not correct.

I also looked and I think the addition of watson.ai in this PR was wrong #1574. The same applies, that is the product name, not a namespace.

Blocking it for now to avoid things getting merged.

[joaopgrassi]

I also believe we should not change this to be a namespace. Vertex AI seems to be the product name.

[gyliu513]

@joaopgrassi watsonx is actually right, please check the following links:

• watsonx.ai https://www.ibm.com/products/watsonx-ai
• watsonx.data https://www.ibm.com/products/watsonx-data
• watsonx.governance https://www.ibm.com/products/watsonx-governance

[joaopgrassi]

Yes, I also checked those. The problem here is that I personally see the dot in the name as part of the brand name. In semconv a dot is to be used for namespacing things, so using the dot just because the brand name uses it is wrong IMO because it clashes with the deep namespace meaning the dot has in semconv.

[joaopgrassi]

Removing my blocker.

I got caught up and passed through me that this is just renaming the enum values of gen_ai.system. So this is fine.

What triggered my concern is that it can get us into iffy territory if we start using the product name "dot" in semconv, as we won't know its meaning.

It is of course fine if the dot is used for namespacing purposes (to add more attributes under it, or sub-namespaces).

[joaopgrassi]

OK I re-read and thought about it, giving this is only changing the enum value all if fine.