Improve memory allocs and speed of encoding maps by using Go 1.18+ features

[dinhxuanvu]

Description

Since go 1.18, the reflect package introduces MapIter.SetKey and MapIter.SetValue that will do fewer memory allocation for map iteration which is frequently used for CBOR encode operation. Plus, usage of sync.Pool will further reduce memory allocation by reusing the shared memory in the pool. Lastly, the Value.SetZero method (available since go 1.20) is helpful to release memory allocation to the GC when is no longer needed.

Signed-off-by: Vu Dinh [email protected]

PR Was Proposed and Welcomed in Currently Open Issue

• This PR was proposed and welcomed by maintainer(s) in issue feature: improve memory allocation for marshall go map type #467
• Closes or Updates Issue feature: improve memory allocation for marshall go map type #467

Checklist (for code PR only, ignore for docs PR)

• Include unit tests that cover the new code
• Pass all unit tests
• Pass all 18 ci linters (golint, gosec, staticcheck, etc.)
• Sign each commit with your real name and email.
  Last line of each commit message should be in this format:
  Signed-off-by: Firstname Lastname [email protected]
• Certify the Developer's Certificate of Origin 1.1
  (see next section).

Certify the Developer's Certificate of Origin 1.1

• By marking this item as completed, I certify
  the Developer Certificate of Origin 1.1.

Developer Certificate of Origin
Version 1.1

Copyright (C) 2004, 2006 The Linux Foundation and its contributors.
660 York Street, Suite 102,
San Francisco, CA 94110 USA

Everyone is permitted to copy and distribute verbatim copies of this
license document, but changing it is not allowed.

Developer's Certificate of Origin 1.1

By making a contribution to this project, I certify that:

(a) The contribution was created in whole or in part by me and I
    have the right to submit it under the open source license
    indicated in the file; or

(b) The contribution is based upon previous work that, to the best
    of my knowledge, is covered under an appropriate open source
    license and I have the right under that license to submit that
    work with modifications, whether created in whole or in part
    by me, under the same open source license (unless I am
    permitted to submit under a different license), as indicated
    in the file; or

(c) The contribution was provided directly to me by some other
    person who certified (a), (b) or (c) and I have not modified
    it.

(d) I understand and agree that this project and the contribution
    are public and that a record of the contribution (including all
    personal information I submit with it, including my sign-off) is
    maintained indefinitely and may be redistributed consistent with
    this project or the open source license(s) involved.

[fxamacker]

@dinhxuanvu Thanks! I'll take a closer look this weekend.

[x448]

@dinhxuanvu @fxamacker The benchmarks shown in the issue look fantastic! The way this PR continues to support for Go 1.17 while using newer features with Go 1.20+ compiler is good.

Unless I'm mistaken, it seems about 300 out of 358 lines added by this PR is existing code that's unmodified by this PR, but it shows up as new code from new contributor. E.g. git blame for encode_map_go117.go.

This PR can benefit from being more clear about what lines of code were modified or remains unmodified code that's already audited/reviewed in past releases. It would help code reviews, troubleshooting/bisecting, and attributing work.

[fxamacker]

@dinhxuanvu Thanks again for opening this PR! 👍 Amazing speed/memory improvements! 🎉 Looking forward to merging this!

This PR looks good! However, the old code required this PR to use too much duplicate code.

Since the duplicated functions are large and complicated, it would best to avoid maintaining multiple versions of them.

Given this, I opened PR #473 to refactor the old code to allow this PR to provide its optimizations by implementing just 2 functions.

See encode_map_go117.go in PR #473 for more details and please let me know if it works for you or if it requires changes.

Thanks again for this PR!

[fxamacker]

@dinhxuanvu sorry about the merge conflict and thanks for reviewing and approving PR #473!

I appreciate your PR and am looking forward to merging this after it's updated! 👍

[dinhxuanvu]

@fxamacker The PR is updated and I did the benchstats again and the result looked very much identical from what I reported in the issue. Thanks for helping with the refactoring! PTAL

[fxamacker]

Great work and thanks for updating this PR!

Not only does this PR improve memory allocs 🗜️ but it also improves speed 🚀 of encoding maps!