stylo: review why these layout/reftests/reftest-sanity font tests are skip-if(styloVsGecko)

RESOLVED WONTFIX

Status

()

Core
CSS Parsing and Computation
RESOLVED WONTFIX
9 months ago
9 months ago

People

(Reporter: cpeterson, Unassigned)

Tracking

57 Branch
Points:
---

Firefox Tracking Flags

(firefox57 affected)

Details

(Reporter)

Description

9 months ago
https://searchfox.org/mozilla-central/search?q=-if%5C(stylo&case=true&regexp=true&path=layout%2Freftests%2Freftest-sanity%2Freftest.list

Why are these font-size tests skipped?

layout/reftests/reftest-sanity/reftest.list

135	ref-pref(font.size.variable.x-western,16) skip-if(styloVsGecko) == font-size-16.html font-default.html
136	ref-pref(font.size.variable.x-western,16) skip-if(styloVsGecko) != font-size-24.html font-default.html
137	ref-pref(font.size.variable.x-western,24) skip-if(styloVsGecko) == font-size-24.html font-default.html
138	ref-pref(font.size.variable.x-western,24) skip-if(styloVsGecko) != font-size-16.html font-default.html
147	ref-pref(font.size.variable.x-western,16) test-pref(font.size.variable.x-western,24) skip-if(styloVsGecko) != font-default.html font-default.html
148	ref-pref(font.size.variable.x-western,24) test-pref(font.size.variable.x-western,16) skip-if(styloVsGecko) != font-default.html font-default.html
149	ref-pref(font.size.variable.x-western,24) test-pref(font.size.variable.x-western,24) skip-if(styloVsGecko) == font-default.html font-default.html
157	fuzzy(1-10,1-250000) fuzzy-if(false,5-10,250000) skip-if(styloVsGecko) == fuzzy.html fuzzy-ref.html
158	fuzzy(0-0,250000) skip-if(styloVsGecko) != fuzzy.html fuzzy-ref.html
159	fuzzy(1,0-2) skip-if(styloVsGecko) != fuzzy.html fuzzy-ref.html
We may not really need to skip them, but in general, I think tests in reftest-sanity doesn't really make much sense for styloVsGecko test, because those are mostly for checking handling of the various annotations.

In styloVsGecko, we have a very special reference page (which is the same page rendered in different engine). I have no idea how reftest harness handles this case, but I don't think it's worth worrying about reftest-sanity for styloVsGecko. I would suggest we should probably just skip the whole directory for styloVsGecko.
I agree with xidorn. As long as these tests are passing in stylo-vs-stylo mode I don't think it's worth investigating, since we're turning off styloVsGecko soon anyway.
Status: NEW → RESOLVED
Last Resolved: 9 months ago
Resolution: --- → WONTFIX
From previous discussion on email, it seems to me we are not turning off styloVsGecko soonish. We may still want to see it when we work on Android support.

But either way, the reftest sanity test wasn't designed for this kind of special mode, and doesn't make sense in many cases. If we want to check sanity for styloVsGecko, we should design an independent suite of sanity test, but that's probably not worth.
You need to log in before you can comment on or make changes to this bug.