feat(github): Don't store updatedAt field in GraphQL cache record

[zharinov]

Changes

• Separate updatedAt and cache data in order to optimize cache I/O
• Remove updatedAt field
• Rewrite cache only if any of items has changed

Context

• Closes: Avoid package cache set every run for GitHub tags/releases #21023

Documentation (please check one with an [x])

• I have updated the documentation, or
• No documentation update is required

How I've tested my work (please select one)

I have verified these changes via:

• Code inspection only, or
• Newly added/modified unit tests, or
• No unit tests but ran on a real repository, or
• Both unit tests + ran on a real repository

[rarkins]

What is the expected outcome?

When I do a dry run on this repo two times in a row I still see a lot of package cache sets:

DEBUG: Repository timing splits (milliseconds) (repository=renovatebot/renovate)
       "splits": {"init": 15133, "extract": 3496, "lookup": 14482},
       "total": 33262
DEBUG: Package cache statistics (repository=renovatebot/renovate)
       "get": {"count": 246, "avgMs": 269, "medianMs": 205, "maxMs": 1643},
       "set": {"count": 24, "avgMs": 431, "medianMs": 354, "maxMs": 1561}
DEBUG: http statistics (repository=renovatebot/renovate)
       "urls": {
         "https://api.github.com/graphql (POST,200)": 26,
         "https://api.github.com/repos/renovatebot/.github/contents/default.json (GET,200)": 1
       },
       "hostStats": {"api.github.com": {"requestCount": 27, "requestAvgMs": 1171, "queueAvgMs": 0}},
       "totalRequests": 27

I assume those 24 are for github tags/releases and that they're unnecessary.

[rarkins]

My expectation is that when I run on a repo two times in a row that there are no package cache sets on the second run.

[zharinov]

That's because we're still saving new updatedAt value to the package cache each run (it costs about 5ms).

[rarkins]

Why do we need to set the updatedAt?

Let's say I asked you not to set it every run. What would we lose?

[zharinov]

Hah, turns out it's the legacy field from the time when we tried to do zero GraphQL queries per run. So the answer is we only lose the potential possibility of doing zero queries, but seems like we don't want it anyway.

[viceice]

🤷‍♂️ don't we have a equal comparison lib already?

[zharinov]

🤦 We do

[viceice]

ok, so please use the existing 😁

[rarkins]

DEBUG: Package cache statistics (repository=renovatebot/renovate)
       "get": {"count": 222, "avgMs": 275, "medianMs": 186, "maxMs": 1669},
       "set": {"count": 0}

[renovate-release]

🎉 This PR is included in version 35.11.0 🎉

The release is available on:

• GitHub release
• 35.11.0

Your semantic-release bot 📦🚀