On 2014-06-05, 2:08 AM, Boris Zbarsky wrote:
On 6/4/14, 11:30 PM, Gavin Sharp wrote:
- benefits to shared API/implementation seem uncontroversial
Agreed.
- specifically, consistency between mochitest/SimpleTest-based
harnesses (mochitest-plain/mochitest-chrome/mochitest-browser) and
xpcshell tests is what we care about primarily. I don't think we
should expand the scope of this discussion beyond that to e.g.
testharness.js.
That works for me.
- I'm not particularly attached to CommonJS assertions, and probably
actually favor using the existing SimpleTest API as "the one true test
assertion API" since it's more familiar to me and seems to be working
well. Sounds like bz/ehsan/sicking all agree
I do.
Assuming agreement across those, that suggests we should modify
Assert.jsm to use the SimpleTest API instead, and continue with this
plan to gradually encourage use of Assert.jsm for the mochitest-based
harnesses and xpcshell.
If it's using the SimpleTest API, we could easily, and should, just
switch the mochitest-based harnesses to it wholesale.
Agreed on all of the above.
(Separately, we could then perhaps address any
shortcomings with the existing SimpleTest assertions
Top of the list: make is() do a === instead of == and get rid of ise().
I'd hope this can actually be done without too many cycles through try...
We only have 237 ise calls, so at least the latter should be fairly
simple. The number of things which would fail if is() does a === is yet
to be known before someone starts trying...
Cheers,
Ehsan
_______________________________________________
dev-platform mailing list
dev-platform@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-platform
