test_borders_table_collapse and assert_different_renderings seem flaky

[benoitkugler]

Hi,

I think there is currently two issues regarding test_borders_table_collapse :

1. Under collapsed borders, the groove and outset styles render the same (as well as the inset/ridge pair), so that the test should not ask for different renderings
2. I think Weasyprint actually properly follows the spec, but that assert_different_renderings seems broken : at tests/draw/init.py line 85, the pixels comparison pixels_1 == pixels_2 seems to always be False, since the variables are compared by reference, not by actual pixel content. A simple fix should be to change the comparison to list(pixels_1) == list(pixels_2)

From what I understand, no change is required to the library, only to the tests.

[liZe]

Hi,

Thanks for the report, I don’t know how you caught this! Interested in providing a pull request?

[benoitkugler]

I'm writing a port of WeasyPrint for the Go programming language. Thanks by the way for your work, it has been very interesting to dig into the depths of HTML rendering !
I've discovered the issue because the Go version of the test was failing, yet with a seemimgly correct rendering.

I can send a PR, not sure when however

[liZe]

I can send a PR, not sure when however

Let me do this for you 😄

Thanks again for the report!

[benoitkugler]

Let me do this for you 😄

Thank you ;)

On est jamais mieux servi que par soi-même 😅