Remove deprecated outputs datadog and kafka

[mstoykov]

No description provided.

[codecov-commenter]

Codecov Report

Merging #2081 (3dc32a7) into master (cc0361c) will decrease coverage by 0.03%.
The diff coverage is 0.00%.

@@            Coverage Diff             @@
##           master    #2081      +/-   ##
==========================================
- Coverage   72.13%   72.10%   -0.04%     
==========================================
  Files         181      180       -1     
  Lines       14289    14243      -46     
==========================================
- Hits        10308    10270      -38     
+ Misses       3353     3348       -5     
+ Partials      628      625       -3     

Flag	Coverage Δ
ubuntu	72.04% <0.00%> (-0.04%)	⬇️
windows	71.72% <0.00%> (-0.07%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
cmd/outputs.go	0.00% <0.00%> (ø)
lib/testutils/minirunner/minirunner.go	78.26% <0.00%> (-2.90%)	⬇️
lib/executor/vu_handle.go	95.32% <0.00%> (+0.93%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update cc0361c...3dc32a7. Read the comment docs.

[na--]

Why do we have additions to go.mod when we're removing a dependency? 😕

[mstoykov]

This now specifies the version that was required by the removed dependency.
As go always get the lowest possible version, when you remove a dependency that was requiring the highest possible version it will now require something different, which is likely not what you want, so it needs to now require that specific version in go.mod in order for the dependency version to not change

[na--]

So right now (and probably before), the versions of these indirect dependencies the k6 vendors are not the ones that were required by the actual direct dependencies' go.mod files, right? If so, it's probably not a big deal, but we should probably clean this up now or when we update the direct dependencies of k6? We don't have a good reason for the current state, right?

[mstoykov]

the versions of the inderect dependencies were probably all added as case like
The general case I think is, we have:

k6 -> A -> C(v1.2.0)
k6 -> B -> C(v1.3.0)

we dropped dependency B, and if this isn't added we will now get C v1.2.0 as that is the highest version required.

So go mod adds an indirect dependency to C v1.3.0, so we don't change the version of something we didn't intend to.
If at a later point we update A and it requires C v1.4.0 it will be that version that will be used, and from my quick test it will also remove the indirect dependency in go.mod(as it's no longer changing what versions we will get).
If we drop A (in this case that would be api2go ;) ) we will probably also drop the indirect dependencies as well.

So we can clean whenever and change the dependency version, which seems like a way to get something broken. Or just when updating/dropping dependency it will fix itself

[mstoykov]

example: do go get google.golang.org/grpc and see how it drops the indirect dependency on xerrors

[imiric]

LGTM, love to see the negative diff stat. :)

I would've preferred to leave this for the end of the v0.34.0 cycle to keep master working until right before release, but it's not a big deal.

[mstoykov]

I do prefer to do it now, so we can actually update kafka extension output to depend on version of k6 without that k6 depending on the kafka extension output

[na--]

Yeah, dropping them ASAP is probably going to be cleaner, and xk6 users shouldn't be affected, since by default they use the latest k6, right? So...