Implement native hash-map compare across all implementations about mal

[kanaka]

I'll go ahead and claim perl too.

[dubek]

I'm out of town till next week so ain't gonna advance in the need few days.
Leave some for me!
On Nov 18, 2015 16:38, "Joel Martin" [email protected] wrote:

I'll go ahead and claim perl too.

—
Reply to this email directly or view it on GitHub
#116 (comment).

[dubek]

I'm claiming ocaml, and afterwards I'll try racket.

[dubek]

ocaml fix in #118

[dubek]

racket fix in #119 .

While coding this, I noticed another problem with implementations that rely on "native" = or equal? operator - they fail when you have nested vector/list comparisons. For example, in Racket, before the fix:

user> (= [1 2] (list 1 2))
true
user> (= [[1 2]] [(list 1 2)])
false

The first one is good (and tested in tests of step4); the second one is incorrect (and now fixed in #119 ).

So besides broken hash-map equality we also have this case. Not sure if you want to add this test case as mandatory (next to the other cases of vector/list equality in step4) or soft-fail.

[kanaka]

It should eventually be with the other vector/list tests, but to avoid breaking Travis until they are fixed, let's make it soft for now until we have most fixed.

[kanaka]

The embedded sequence implementations is tracked here: #123

[kanaka]

Just pushed a fix for R.

[dubek]

Guile fix in #124.

[kanaka]

matlab pushed.

[ivern]

@kanaka want me to fix the Kotlin implementation?

[kanaka]

@ivern that would be great. FYI, the simplest method I've found is to compare the lengths (or the keys if the native impl doesn't have a length method/function) and then iterate through all the keys from one map, making sure that each key both exists in the second and that has the same value.

[ivern]

I've fixed my pull request to contain a single commit.

[ivern]

Kotlin should be updated to fixed now that it's merged.

[kanaka]

I merged before the travis was finish with the PR, and unfortunately, travis doesn't pull the right thing so tests start to fail after the merge. So I'll wait for the travis build for the merge itself (https://travis-ci.org/kanaka/mal/builds/104935848) to finish and verify and then close this if everything checks out.

[kanaka]

Looks good. Closing.