630117 - rename typed array slice() to subarray()

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

We decided last week that we screwed up with the slice() naming, and changed the name of that function to subset().  We should get this change in, so that we can get a real slice() in the future.  (Current slice() creates a new view object, but still references the same underlying data.)  Marking this as a softblocker for API consistency, but I'll post a patch tomorrow.

Comment 1

[Brendan Eich [:brendan]]

Why "subset"? There's no sane set-theoretic interpretation. As P. Tucker Withington pointed out, the view objects are like Common Lisp's "Displaced Arrays". Not that you should use that name. "sliceView"?

/be

Comment 2

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Attachment: rename slice -> subset

renamed, and tests updated; also added a test to ensure that subset() behaves as intended (references same buffer).

Comment 3

[David Humphrey (:humph)]

CC'ing some others who might get broken by this.  FYI, gentlemen.

Comment 4

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

I don't have a good answer as to "why subset", other than that Ken & I liked that name, and it didn't collide with "slice"; the slice wart we should've caught a while ago (and I think Brendan even mentioned it originally :/)

Comment 5

[Dave Herman [:dherman]]

Wow, "subset" smells pretty funky to me... neither the original array nor the result array is a set. I can see how "slice" could be misleading given that array.slice produces a fresh array. How about "subarray"?

Some of the original work on this kind of API was known as "subsequences" since it was about taking pieces of (possibly mutable) sequence types that share references:

    http://portal.acm.org/citation.cfm?id=133234

Seems like a good analogy: subarray : array :: subsequence : sequence. Especially since an array is a kind of sequence.

Dave

Comment 6

[Jeff Walden [:Waldo]]

subrange, subview, sharedRange, sharedSlice

Comment 7

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

I don't think we get to pick a new name at this point... Ken already made the change in WebKit.  Cc'ing him here -- really should get some JS folks involved in the typed array work :)

Comment 8

[Jeff Walden [:Waldo]]

Fair enough, throwing them out just in case.

As far as JS people getting involved, my understanding is that ECMA is doing its own thing in this space, looking to supplant typed arrays.  See comment 575688 comment 6.

Comment 9

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Comment on attachment 508511 [details] [diff] [review]
rename slice -> subset

(Argh, I hate that you're not :waldo :-)

Yeah, TC39 is doing its own thing, but we have this now and will likely have it for quite a while... we're already depending on it very heavily for a lot of Fx4 launch demos and the like :-)

Comment 10

[Jeff Walden [:Waldo]]

Comment on attachment 508511 [details] [diff] [review]
rename slice -> subset

rs=me assuming it builds and tests out fine.

We should eventually move this over to use the atom stuff and shared const char[] in jsatom.h, but that certainly doesn't need to happen now, nor should it, really.

Comment 11

[Kenneth Russell]

Since we've already broken content with the slice -> subset renaming, I suppose if we quickly renamed it again it wouldn't be that bad.

Of the above suggestions, I like subarray the best. There are more view types of ArrayBuffers than just the Typed Arrays, in particular DataView, which doesn't support this operation. Therefore subview doesn't seem appropriate. subrange seems okay, but if we're adding more letters to the method name it seems to me we might as well keep the name focused on the array-like nature of these objects.

Vlad, what do you think?

Comment 12

[Grant Galitz]

"Since we've already broken content with the slice -> subset renaming, I suppose
if we quickly renamed it again it wouldn't be that bad." Let's go 'derper' with the renaming now: http://images0.memegenerator.net/ImageMacro/3784263/we-need-to-go-derper.jpg?imageSize=Medium&generatorName=Inception

Comment 13

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

If you're ok with an additional rename, subarray sounds fine, or 'subarrayview', but now maybe we're trying too hard.

Comment 14

[Brendan Eich [:brendan]]

+1 on subarray.

(subarrayview is overlong and violates camelCaps -- subarray does not if it's one word, which neologisms and back-formations quickly become, esp. in hacker jargon.)

/be

Comment 15

[Kenneth Russell]

If we're in agreement on subarray, I'll update the Typed Array spec and conformance tests and do yet another rename on the WebKit side. Vlad?

Comment 16

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

sounds good to me, i'll update this patch

Comment 17

[Grant Galitz]

ಠ_ಠ

Comment 18

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Attachment: rename slice -> subarray

slice -> subarray!

Comment 19

[Grant Galitz]

I hope I don't come back here to see that it's been changed another two times.

Comment 20

[Jeff Walden [:Waldo]]

Comment on attachment 508837 [details] [diff] [review]
rename slice -> subarray

rs=me again

Comment 21

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

http://hg.mozilla.org/mozilla-central/rev/84c8b33619e0

death to slice, long live subarray.

Comment 22

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

argh, the commit comment is wrong; I meant subarray! (code is correct)

Comment 23

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

This turned tests orange. I checked in a attempted fix:
http://hg.mozilla.org/mozilla-central/rev/808970eca6d5

Comment 24

[Brendan Eich [:brendan]]

(In reply to comment #23)
> This turned tests orange. I checked in a attempted fix:
> http://hg.mozilla.org/mozilla-central/rev/808970eca6d5

And the checkin comment/stale bug summary bites. Sicking renamed the calls from slice to subset!

Vlad, can you fix? rs=me

/be

Comment 25

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Ugh, didn't realize that the checkin-comment didn't reflect reality. Backed out the whole thing since this has already kept the tree orange for several cycles and my fix was risky enough as I didn't have time to compile a fresh tree and test.

http://hg.mozilla.org/mozilla-central/rev/298c555d0c6e

Comment 26

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

For what it's worth, my patch didn't apparently cover all cases which needed s/slice/subfoo/

Comment 27

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Yeah, sorry -- I tested JS, but forgot that the webgl tests had coverage of this :(  I have everything patched locally (with correct commit message this time), but refreshing my build to make sure everything's good.

Comment 28

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

Attachment: update webgl tests

The additional patch.

Comment 29

[Vladimir Vukicevic [:vlad] [:vladv] (needinfo me, slow to respond)]

relanded with additional webgl test fixes.

Comment 30

[Eric Shepherd [:sheppy]]

Reminder that this is one that should have gotten a dev-doc-needed keyword on it. :)

That said, documentation has been updated, thanks to Dave Humphrey's blog post catching my eye.