Guillaume Ayoub

(Note that I’ve also tried to fix #2055, but not pushed my code yet. I’ll clean my code and push it soon.)

I’ve open #2058 so that you can see what I’ve managed to get. It’s far from perfect, and as explained above I’m not sure that we can easily get a...

> Appreciate if you can point out available test cases that fail it or some reference so that relevant test cases can be written. Here’s an example: ```html abc ```...

Thanks for your work on this. Could you please avoid to change `draw.py` and `boxes.py`? What we want is to fix the layout. Changing what `x` means is a workaround,...

> This is not noticed in pdf as I think it's rendered in "vector" but when rendered as pixel, the one pixel difference, bounded in red, can be seen when...

I have finally found a solution that is cleaner from my point of view. Both the tree building and the drawing parts are now independent from `direction`, the rtl logic...

> From the above x values, cells are still being drawn from right-to-left, thus the one pixel difference problem mentioned in [#2056 (comment)](https://github.com/Kozea/WeasyPrint/pull/2056#issuecomment-1941522665) still persists. As explained above, trying to...

> But if the objective is to always draw left-to-right regardless of the `direction` (which I see as one way to ease the complication), then I wish to highlight that...

Thanks for the report. Looks like it’s a bug in WeasyPrint, even if the specification is not 100% obvious about this (talking about ["string value"](https://drafts.csswg.org/css-gcpm/#content-function-header). Related to #137.

> I'm working on a project which would really benefit from this, would be cool to see it implemented Yes, that would be great. I think that it’s possible by...