Michał Siedlaczek

We've upgraded the docs since this was opened, here's one link of interest: https://pisa.readthedocs.io/en/latest/experiments/regression-robust04.html

I'm closing this in favor of #521

This doesn't seem to show up anymore.

No longer seeing any errors in our code.

Alternatively, here's the diff for "i" aligned with "l": ``` diff --git a/source/Hack-Regular.ufo/glyphs/i.glif b/source/Hack-Regular.ufo/glyphs/i.glif index dbbe950bc..592ff10a4 100644 --- a/source/Hack-Regular.ufo/glyphs/i.glif +++ b/source/Hack-Regular.ufo/glyphs/i.glif @@ -4,38 +4,38 @@ - - - - -...

Here's a preview for the i-shifted-to-align-with-l change (starts with 19px and goes down by one pixel to 10px):  

I would like to see this feature as well

This is a tricky issue because of efficiency considerations. But if the problem is the stability of of the final sort, how about we add this additional sort condition only...

BTW, I have this PR pending: https://github.com/pisa-engine/pisa/pull/507 where I implement a custom pop-push operation instead of using `std::pop_heap`. We could also implement our own `pop_push` (used only before the heap...

@seanmacavaney FYI, I opned a PR for this, if you'd like to give it a look.