Re: [community-group] First-Class Mode Support with `$modes` Array (#348)

> Surely, despite the misunderstanding, the original issue exists. "the spec currently lacks a standardized mechanism for parsers to programmatically discover which sibling keys represent variants", or to put it another way, I have no reliable mechanism to programmatically determine that 'accent' has N number of definitions - which should be of some concern to those seeking to transform the design.tokens into platform code via StyleDictionary.

This is not accurate. The resolver module does solve this problem completely. I do disagree with the “this solves a different problem and is complementary” assertion. The resolver module _does_ solve single-mode problems just as well as multi-modal intersections, so the simplifications are incorrect. [There’s a resolver playground](https://dtcg-resolver-playground.pages.dev) that may help understanding (more guides TBD).

When you have tokens coexisting in the same files with different IDs, the expected behavior is all these tokens may be used in the same context all together, i.e. are NOT variants ([longer explanation](https://terrazzo.app/docs/guides/modes/), applicable to any tool). Put another way, if you conditionally only use some tokens in some contexts, and other tokens in alternate contexts, you’re actually using multiple design systems at once, which is more unwieldy (and error prone) than using a proper mechanism to treat the same token ID as having the same purpose in the system, and relying on smarter mechanisms to contextually change the value.

> which should be of some concern to those seeking to transform the design.tokens into platform code via StyleDictionary.

I think you’re raising another problem, which is that [Style Dictionary doesn’t support the 2025.10 version](https://github.com/style-dictionary/style-dictionary/issues/1590) which I’m trying to help them with. I heard that the team will start investigating the impact later this month, timeline TBD. So the tooling is lagging behind the specification somewhat, but the end result of declaring alternate values with a resolver file is sufficient to solve all the problems you raised. I think we just may need more education and guides on how to achieve all that! I fully recognize it’s a complex thing to grok, especially when tools don’t allow experimenting at the moment.

-- 
GitHub Notification of comment by drwpow
Please view or discuss this issue at https://github.com/design-tokens/community-group/issues/348#issuecomment-3512137143 using your GitHub account


-- 
Sent via github-notify-ml as configured in https://github.com/w3c/github-notify-ml-config

Received on Monday, 10 November 2025 14:44:48 UTC