Encoding: encodeInto()

[annevk]

For whatwg/encoding#166.

[annevk]

I found it a little hard to write tests without an implementation, but I think this framework is adequate for any number of tests we'd want to write. Probably at this point it's most helpful to review that framework and then when implementations start we can fill out the number of tests?

[ricea]

Looks good. I think this is a good framework to test encodeInto(). I haven't actually checked the byte values, I'm assuming they're correct.

Edit: I think we will need a separate test that bytes after the last written codepoint are not modified. With this test framework we can't tell whether they're left alone or overwritten with null bytes.

[hsivonen]

Looks good to me. Thanks.

To test an edge case that's hard for encoding_rs, it would be good to add a test whose output buffer size is 4 and whose input is an ASCII character followed by a snowman (or other 3-bytes-as-UTF-8 character).

Also: Output buffer size 4 and input contains two 2-bytes-as-UTF-8 characters.

[inexorabletash]

Mentioned on the spec PR but:

Needs cases where the destination view is offset/subset of the backing array buffer, e.g.

const ab = new ArrayBuffer(5);
const view = new Uint8Array(ab, 1, 2);
new TextEncoder().encodeInto("ABCDE", view);

[annevk]

I wrote tests for more complicated view setups which was a lot of fun, but an implementation would be good at this point as I'm worried for off-by-one errors.

(I also added Henri's tests. I think I covered everything now.)

[annevk]

I've written all the various tests requested thus far, I think, including some additional ones I thought of myself.

[foolip]

FYI, I've rebased this branch to trigger "Travis CI - Pull Request", manually setting things right after the travis-ci.com transition in #14499 changed the name of the required check, which applies retroactively.

[annevk]

Does someone want to review the additions?

[hsivonen]

@foolip, the chrome[experimental] and firefox[experimental] tasks started 20 days ago but don't show up as finished.

[foolip]

@lukebjerring can you take a look at what's going on with the wpt.fyi checks? I guess this is web-platform-tests/wpt.fyi#950.

[ricea]

I wrote a slow and dumb polyfill. With the change I noted the tests pass.

[lukebjerring]

Yep, attempting to update says:

Could not find runs to compare: no test run found for firefox @ 60bfa1e with label pr_base

[annevk]

I plan to land these tomorrow, together with the change to the Encoding Standard.

Thanks for the review @ricea!

[lukebjerring]

Looks like @Hexcles has succeeded in uploading empty reports 🎉
Now we can easily see the outcome is CRASH in Chrome and TIMEOUT in Safari.
https://staging.wpt.fyi/results/?pr=14505

Please fix that before this lands.

[annevk]

@lukebjerring can you reproduce it? I can't.