Use Buffers for cache records

[zharinov]

Describe the proposed change(s).

Encoding to base64 was a bad idea, because it slows down cache writes about twice.
Instead, we have to do JSON.stringify, compress it into a Buffer and then pass it directly to Redis or Cacache.

[zharinov]

Since new format is incompatible, we need to use different cache keys.
In order to reuse older cache records until they're expired, let's fallback to the older cache key during cache get operations.

[rarkins]

We can't detect it based on the payload or some other heuristic?

[viceice]

don't we've a cache version? 🤔 we previously used that.

can we use some magic bytes at buffer start to detect version?

[zharinov]

don't we've a cache version? 🤔 we previously used that.

It's repository cache concept and it is located inside JSON which is not yet decoded at the moment of Buffer vs base64 decision

[viceice]

what if we simply try to decode the new way and fallback to old way when failed?

[renovate-release]

🎉 This issue has been resolved in version 37.421.8 🎉

The release is available on:

• GitHub release
• npm package (@latest dist-tag)
• 37.421.8

Your semantic-release bot 📦🚀