misc(ci): increase fetch-depth for codecov SHA

[brendankenny]

Possible fix for codecov issues. It seems to have some problem with the checkout fetch-depth, but 100 seems to fit what it's asking for and codecov/codecov-action#190 has been closed for over a month now.

No reason it needs to go rooting around though, we already know the commit hash.

codecov issue screenshot

[paulirish]

in basics we get good depth, but not in the unit jobs. there we get the default depth=1.

i'm fine with this as a fix (attempt), but.. if not.. there's a decent chance my lovely git-get-shared-history script would do the work to satisfy this as well. :)

[patrickhulce]

If the action fixed it maybe we should just update to the latest? We're pinned to 6004246 right now

[brendankenny]

If the action fixed it maybe we should just update to the latest? We're pinned to 6004246 right now

I can update but from that conversation it's fixed in the bash script, and we're on Bash-20210129-7c25fce, well past where the fix supposedly landed.

[patrickhulce]

proof is in the pudding though, if this passes codecov checks I'll LGTM too :)

[patrickhulce]

we're on Bash-20210129-7c25fce, well past where the fix supposedly landed.

Wait, so our fixed action commit just updates itself and runs whatever the hell it wants? 😆

[brendankenny]

Warning: Unexpected input(s) 'override_commit', valid inputs are ['name', 'token', 'file', 'files', 'directory', 'flags', 'path_to_write_report', 'env_vars', 'fail_ci_if_error']

haha, see if you can see the problem with these docs: https://github.com/codecov/codecov-action#arguments

Codecov's Action currently supports five inputs from the user...

Table with 30 options...

[patrickhulce]

fetch-depth: 2 LGTM assuming it works :)

(seems like it did? https://github.com/GoogleChrome/lighthouse/pull/12079/checks?check_run_id=1875496980)

[brendankenny]

in basics we get good depth, but not in the unit jobs. there we get the default depth=1.

ah, yeah, I didn't notice that. I'll try deeper.

i'm fine with this as a fix (attempt), but.. if not.. there's a decent chance my lovely git-get-shared-history script would do the work to satisfy this as well. :)

Honestly I don't know if depth matters all that much for CI perf...between some level of caching and overhead, there's barely a difference between git clone in basics (depth 100) and unit (depth 1) and then we spend like 10x that time building LR right afterwards :)