feat: lazy initialization of the gzip stream

[fungiboletus]

Which problem is this PR solving?

A gzip stream is created by opentelemetry-exporter-otlp-http even when it's never used. Moreover, it's also never ended. It's particularly visible when using the testing framework Jest, that will complain about it with the following message : Jest has detected the following 1 open handle potentially keeping Jest from exiting: ZLIB.

Short description of the changes

Having a way to call gzip.end() would fix the issue, but I think it's first preferable to avoid creating the gzip stream in the first place.

Type of change

Please delete options that are not relevant.

• Bug fix (non-breaking change which fixes an issue)

How Has This Been Tested?

I ran npm test in the root folder and experimental folder.

Checklist:

• Followed the style guidelines of this project
  - [ ] Unit tests have been added (not required IMHO)
  - [ ] Documentation has been updated (not required)

[linux-foundation-easycla]

The committers are authorized under a signed CLA.

• ✅ Antoine Pultier (e1ff0fd)

[codecov]

Codecov Report

Merging #2581 (4bb0158) into main (1abda14) will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##             main    #2581   +/-   ##
=======================================
  Coverage   93.00%   93.00%           
=======================================
  Files         138      138           
  Lines        5092     5092           
  Branches     1095     1095           
=======================================
  Hits         4736     4736           
  Misses        356      356           

[dyladan]

We should add end() to the shutdown method of the exporter as well. Can you please add a todo comment so we don't forget?

[dyladan]

Please fix the conflicts

[fungiboletus]

Conflicts are fixed and I added the comment. I'm not sure how you do about the merge commit because I can't see any in the git log of the main branch, but perhaps you rebase everytime.

[dyladan]

All PRs are squashed when we merge. That's why they have to be up to date. Speaking of which, this PR is out of date so it needs to be updated in order to be mergeable.

[dyladan]

Re-running tests to see if failure is persistent.

[dyladan]

Looks like the error wasn't caused by this PR. Please update the branch so it can be merged.

[dyladan]

Please update the branch so this can be merged. I want to cut a release and I'd like this to be in it.

[fungiboletus]

I'm working on it now.

[fungiboletus]

Is the merge enough or should I do something else ?

[dyladan]

No thats fine as long as the tests pass