## Summary
Decision around CSS architecture for the shared library. Topic encompasses: CSS preprocessor (Less, SassLESS, CSS variablesSass), conventions (BEM, utility classes, etc), design tokens, how styles should be exported, etc.
## Considerations
### Critical to unblock now
#### CSS preprocessor
* WVUI,1. other WMF team-specific libraries**LESS**: WVUI, and MediaWiki use Lessother WMF team-specific libraries, while WiKit uses Sass for reasons documented in [[ https://phabricator.wikimedia.org/T286951#7233145 | this comment ]]and MediaWiki use LESS. [[ https://www.mediawiki.org/wiki/Requests_for_comment/LESS#Why_LESS?_Why_not_SCSS_/_Sass? | Historically (in 2013), the decision ]] was driven by better PHP implementation of LESS.
===== CSS Postprocessor or native
2. **Sass**: WiKit uses Sass for reasons documented in [[ https://phabricator.wikimedia.org/T286951#7233145 | this comment ]].
**CSS Postprocessor or native**
* Note that CSS variables, a pure CSS option, are not supported in IE11 – see also T286947 for shared components lib and T285045 (for Wikit).
* The other big wants in the room are mixins (CSS [[ https://caniuse.com/?search=%40apply | @apply feature ]]) and nesting (coming to a browser you trust with [[ https://drafts.csswg.org/css-nesting-1/ | CSS Nesting module ]]), which for now and the foreseeable future are not possible without a PostCSS plugin. And a build step.
Under these preconditions, **neither a postprocessor nor native CSS seem to be a viable option for our environment right now**.
#### Design tokens
See T266688 for details and previous discussion, including [[ https://phabricator.wikimedia.org/T266688#6884221 | WMDE's evaluation of using JSON-based tokens ]].
---
### Important but less urgent
#### CSS Utility classes
See T285880 for previous discussion and more links.
#### Style scoping
See T258774 for details on how WMDE approached this for WiKit.
#### Relevant links
---
## Decisions
#### Which CSS preprocessor will we use?
#### What design token architecture will we implement?
#### How will we follow up on important but less urgent topics?
####