Tests for protein-translation don't match its README

[jgallagher]

The README for this exercise describe the problem as (roughly) "given a sequence of codons, output the sequence of proteins". However, the tests in protein-translation (a) only seem to expect a single protein name from a single codon and (b) talk about codon "compression" which is not mentioned in the README; e.g.,

fn test_cysteine_tgy() { // "compressed" name for TGT and TGC
    let info = proteins::parse(make_pairs());
    assert_eq!(info.name_for("TGT"), info.name_for("TGY"));
    assert_eq!(info.name_for("TGC"), info.name_for("TGY"));
}

...

fn test_arginine_name() {
    // In arginine CGA can be "compresed" both as CGN and as MGR
    let info = proteins::parse(make_pairs());
    assert_eq!(info.name_for("CGA"), Ok("arginine"));
    assert_eq!(info.name_for("CGN"), Ok("arginine"));
    assert_eq!(info.name_for("MGR"), Ok("arginine"));
}

I glanced at a couple other languages' versions of this exercise and saw tests that look much more like the README and don't talk about compressed names (go, elixir).

[petertseng]

The questions arising from this:

• So does that mean The protein-translation and nucleotide-codons exercises are the same problem-specifications#268 was incorrect? the nucleotide-codons and protein-translation exercises are not actually the same?
• Should replace nucleotide-codons with protein-translation #258 be reverted? Should the nucleotide-codons exercise be un-deprecated in the Rust track, if it is in fact different from protein-translation?
• Should nucleotide-codons: deprecate in favour of protein-translation problem-specifications#532 be reverted? Should the nucleotide-codons exercise be un-deprecated globally, if it is in fact different from protein-translation?
• Should the tests of protein-translation of the Rust track be changed to match the README?
• How can we to decide what the tests should be? We need someone to work on protein-translation: Implement canonical-data.json problem-specifications#577
• Should the protein-translation readme be changed to include anything about compression?

[jgallagher]

It appears to me that exercism/problem-specifications#268 is correct at a high level:

both these exercises do the exact same thing: they translate codons to their correct names.

but "exact" isn't quite right since the two exercises do so quite differently:

• README for protein-translation problem-specifications#37 added the description that is currently the README of protein-translation - converting a sequence of codons into a sequence of names
• Add nucleotide-codons exercise problem-specifications#93 added a description of what the current Rust protein-translation tests check for - converting an individual codon (possibly compressed) into its singular correct name

I would cast a vote for changing the protein-translation Rust track tests to match the README (and what the protein-translation tracks in other languages check for). I have no opinion about reintroducing nucleotide-codons.

[petertseng]

Thanks for the initial report and the subsequent evaluation!

I also wish to express no opinion on nucleotide-codons, but I will leave on a note for whoever decides to work on exercism/problem-specifications#577 that they may consider whether to include the old tests of nucleotide-codons. At this point we may consider the discussion of nucleotide-codons deferred to that issue.

For the Rust track, I would also agree that the protein-translation tests should change to match the readme.