Bugzilla

(In reply to Daniel Holbert [:dholbert] from comment #4)
> Note that one of the tests in part 3 fails; I've got it annotated as expected-fail for now. I plan to file a followup for that which I can include in the .ini file; toggling ni=me as a reminder.  (Going AFK for a bit right now so I can't do it at the moment.)

I figured out what was going on that was causing that test^ to fail. There were two weird things going on there:
(A) the test was firing its `test()` function (and modifying/removing its attributes) multiple times, since it had multiple loads.
(B) With those unintentional repeated test() invocations, we happened to trigger a subtle interop difference where `removeAttribute('href')` does different things in Firefox vs. Chromium/WebKit, depending on whether xlink:href is present and whether it's been modified.

I've addressed (A) by nerfing repeated calls to the load-handler in the tests associated this bug that were susceptible to this (i.e. the tests where the load-handler kicks off a new image load internally and hence results in itself getting called again).  This makes the test pass now.

I've filed Bug 1909028 to document & hopefully investigate the difference that I described as (B) here.

(In reply to Daniel Holbert [:dholbert] from comment #4)
> Note that one of the tests in part 3 fails; I've got it annotated as expected-fail for now. I plan to file a followup for that which I can include in the .ini file; toggling ni=me as a reminder.  (Going AFK for a bit right now so I can't do it at the moment.)

I figured out what was going on that was causing that test^ to fail. There were two weird things going on there:
(A) the test was firing its `test()` function (and modifying/removing its attributes) multiple times, since it had multiple loads.
(B) With those unintentional repeated test() invocations, we happened to trigger a subtle interop difference where `removeAttribute('href')` does different things in Firefox vs. Chromium/WebKit, depending on whether xlink:href is present and whether it's been modified.

I've addressed (A) by nerfing repeated calls to the load-handler in the tests associated this bug that were susceptible to this (i.e. the tests where the `<image>` element's load-handler kicks off a new image load internally which results in another call to the load handler).  This makes the test pass now.

I've filed Bug 1909028 to document & hopefully investigate the difference that I described as (B) here.

(In reply to Daniel Holbert [:dholbert] from comment #4)
> Note that one of the tests in part 3 fails; I've got it annotated as expected-fail for now. I plan to file a followup for that which I can include in the .ini file; toggling ni=me as a reminder.  (Going AFK for a bit right now so I can't do it at the moment.)

I figured out what was going on that was causing that test^ to fail. There were two weird things going on there:
(A) the test was firing its `test()` function (and modifying/removing its attributes) multiple times, since it had multiple loads.
(B) With those unintentional repeated test() invocations, we happened to trigger a subtle interop difference where `removeAttribute('href')` does different things in Firefox vs. Chromium/WebKit, depending on whether xlink:href is present and whether it's been modified.

I've addressed (A) by nerfing repeated calls to the load-handler in the tests associated this bug that were susceptible to this (i.e. the tests where the `<image>` element's load-handler kicks off a new image load internally which results in another call to the load handler).  This nerfing makes the test pass now.

I've filed Bug 1909028 to document & hopefully investigate the difference that I described as (B) here.