feat: remove zeroed vector by pushing front

[ChrisCho-H]

Remove useless zeroed vector following #3131.

1. Init vector with capacity not to realloc when pushing.
2. If vector is empty(the first iteration), push data from the front.
3. Iterate vector from the front, and copy it from the end.

[coveralls]

Pull Request Test Coverage Report for Build 10559092245

Warning: This coverage report may be inaccurate.

This pull request's base commit is no longer the HEAD commit of its target branch. This means it includes changes from outside the original pull request, including, potentially, unrelated coverage changes.

• For more information on this, see Tracking coverage changes with pull request builds.
• To avoid this issue with future PRs, see these Recommended CI Configurations.
• For a quick fix, rebase this PR at GitHub. Your next report should be accurate.

Details

• 12 of 12 (100.0%) changed or added relevant lines in 1 file are covered.
• 406 unchanged lines in 10 files lost coverage.
• Overall coverage increased (+0.01%) to 82.758%

---

Files with Coverage Reduction	New Missed Lines	%
hashes/src/siphash24.rs	4	97.89%
bitcoin/src/psbt/serialize.rs	6	96.21%
units/src/fee_rate.rs	13	82.02%
hashes/src/internal_macros.rs	16	65.17%
io/src/lib.rs	23	69.39%
bitcoin/src/blockdata/script/owned.rs	31	68.33%
bitcoin/src/crypto/ecdsa.rs	44	46.73%
io/src/bridge.rs	50	0.0%
bitcoin/src/blockdata/transaction.rs	91	89.7%
units/src/amount.rs	128	86.56%

Totals
Change from base Build 10527005040:	0.01%
Covered Lines:	19645
Relevant Lines:	23738

---

💛 - Coveralls

[jamillambert]

The CI fail is because the vec! macro is now only used in the #[test] module and the unused lint does not look in test. If you change line 21 from #[macro_use] to #[cfg_attr(test, macro_use)] it should fix the issue.

[Kixunil]

Nice!

@jamillambert I think use alloc::vec; in tests module should work too and I'd find it nicer. But it's fine to use attribute too.

[tcharding]

@jamillambert debugging PRs like a boss, way to go bro, this is super useful.

[ChrisCho-H]

Thanks for the reviews! I've added use alloc::vec; in tests module.

[apoelstra]

ACK 441aac0 successfully ran local tests

[jamillambert]

ACK 441aac0

[Kixunil]

Needs squashing. I have no idea how @apoelstra got to run his per-commit tests without catching this.

[apoelstra]

Which commit and cargo command fails? I can run cargo test, cargo test --all-features --all-targets, and cargo test --no-default-features --all-targets on each commit. (Using git rebase -x master which isn't quite correct, but that's my habitual "check each commit manually" command.)

My local tests do a more extensive feature matrix (IIRC it tries each individual feature by itself, and also two random subsets of features, where "random" is seeded by the commit ID or something) but it may have missed something.

[Kixunil]

Oh, it's actually warnings, not errors. If you used -D warnings it'd fail but that is maybe too strict for you. If so then I'll ACK this commit even though I would've liked squashing better.

[apoelstra]

If you used -D warnings it'd fail but that is maybe too strict for you.

I've had it on in the past but I've always felt bad nitting new contributors because individual commits don't pass a super strict level of scrutiny, which wouldn't be needed just for bisection to work.. For a similar reason I only run clippy on the final commit of PRs. (Though there I run it with -D warnings, which also picks up compiler warnings.)

It also worses the "sometimes you need to squash unrelated stuff together to make the compiler happy" problem that we sometimes encounter in Rust (and which we did encounter in the final commit of this PR.)

So my ACK stands.

[Kixunil]

Yeah, I thought so. Might be worth to run only some very limited categories of clippy on every commit (basically those that detect a problem with pretty much 100% certainty).

[Kixunil]

ACK 441aac0