Re:Ladybird: A new cross-platform browser project
by coldtea on Monday September 12, 2022 @03:39PM
(#32815707)
Sure, but I think you have a misconception: that those are like compliance tests for developers or have that role.
Rather, those acid tests were tests that test the rendering engine (in multiple aspects at the same time), and are meant for the end users: to give them a nice visual representation of the browser's progress, or lack thereof, and to push browser developers to fix issues that cause visual glitches (with features drawing the test image selected to be nice for layout writers to have) and have them race each other to pass them...
Re:Ladybird: A new cross-platform browser project
by andrewmcwatters on Monday September 12, 2022 @05:00PM
(#32816896)
I don't have any misconceptions about the state of web technology testing, or how WaSP made an impact on the miserable state of the industry at the time. I work on browser technology.
Re:Ladybird: A new cross-platform browser project
by tannhaeuser on Monday September 12, 2022 @05:26PM
(#32817228)
> there are zero test suites you can use for reference comparison.
What about W3C's web-platform-tests suite [1]?
[1]: https://github.com/web-platform-tests/wpt/tree/master/css
Re:Ladybird: A new cross-platform browser project
by dillondoyle on Tuesday September 13, 2022 @12:45AM
(#32820725)
To test white space, quick thought have not tried, perhaps render some text that hits a chunk of the differences. Start default/normal. like perhaps text has a lot of tabs. JS measure offsetWidth. Or maybe use canvas export image see if base64 is same
Do it again with different CSS settings.
I'm talking about easy stuff, too, like normal flow, margin, border, padding box alignment.
The existing official test suites are entirely manual.
Edit: What you're describing requires you to make the entire world first, then finally test something.
Here's an example: I want to test the white-space property. How do I test that? Oh also, font rasterizers are all different and standards don't dictate what happens after box layout. It's acceptable that glyphs render with different dimensions.
How do I test that? A casual HN reader isn't going to care. Someone actually writing this stuff, will.