This is a meta-bug to track our support for css-grid-2: https://drafts.csswg.org/css-grid-2/ (editor's draft) (in particular since the "subgrid" feature has been moved from the css-grid-1 CR to css-grid-2 to allow more consideration of feature changes in how subgrid is designed and implemented) This bug will be marked as "Depends on" bugs that are needed to complete implementation of css-grid-2. The dependency tree can be used to view a list of these dependencies. Please do not add comments about specific issues to this bug; they belong in their own bugs.

Note: CSSWG decision to separate subgrid to css-grid-2: https://github.com/w3c/csswg-drafts/issues/958#issuecomment-286813071

Please reach out / let me know how I can help. It seems there's need to figure out better what subgrid needs to do, and what it doesn't need to do. I've got ideas. I'm also not sure how to even begin to help figure it out, since I don't know how the render engine operates. I'm also happy to put any questions to the group of authors I've gathered in Slack. It seems we are at a bit of a chicken & an egg. Authors don't know how the rendering engine works, so we can't clearly see what to ask for / how to imagine this. The spec writers are saying, we don't know what authors need. And implementors are saying, the spec isn't finished, so we can't work on this. But authors can't articulate what we want, because we don't know what the implementers know. I'd love to help interrupt this chicken and egg. What is it authors can provide? I've already got a half a dozen working examples of what's needed / broken demos waiting for subgrid. I could churn out more, and ask for others. Would that help?

Jen, per https://github.com/w3c/csswg-drafts/issues/958 I think the main point is that a suboptimal subgrid was being rushed out the door in the CR (perhaps due to various pressures to just ship *something* subgrid), and that the previous (perhaps more challenging but more useful) design iteration with independent axes could use more attempt at implementing instead of being stuck with the rough simplified unimplemented draft in Grid Level 1. I don't think this is a chicken & egg situation, nor one of "spec writers are saying, we don't know what authors need", but one where the aspirational wishes (we wish we could have grid & subgrid at the same time) did not match up with implementation realities (several interop implementations of grid, zero implementations of subgrid), and it makes no sense to keep delaying the stable in deference to the hopeful. Thus decoupling is a sensible path forward that allows Grid Level 1 to proceed through CR, while giving implementers more time to attempt to do a better (read: not spec-rushed) job with subgrid.