You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
{{ message }}
Repository navigation
(RFC) An additional minify tier that reconstructs CPU functionality on demand using tseynit #3236
( ◦°^°◦) 🏴☠️ Arr, I searched fer duplicate RFCs, found nary a one, and spied that this here draft still be missin' its cargo.
Duplicates: none found. The closest related work is #2431 ("Minify shaders in production"), which covers renaming and whitespace minification, but not reconstructing CPU functionality on demand via tseynit.
Missing context: the body only contains the "This is a draft RFC" note, so the proposal cannot be evaluated yet. To make it understandable, it would help to add:
the motivation and concrete use case (why CPU functionality needs to be reconstructed rather than kept from the original build);
what "on demand" means in practice (which call sites, e.g. tgpu.resolve, kernels, or plugin output);
how the tseynit round-trip would work and its tradeoffs (bundle size, startup cost, correctness);
Activity
pullfrog commentedon Oct 21, 2026
( ◦°^°◦) 🏴☠️ Arr, I searched fer duplicate RFCs, found nary a one, and spied that this here draft still be missin' its cargo.
Duplicates: none found. The closest related work is #2431 ("Minify shaders in production"), which covers renaming and whitespace minification, but not reconstructing CPU functionality on demand via
tseynit.Missing context: the body only contains the "This is a draft RFC" note, so the proposal cannot be evaluated yet. To make it understandable, it would help to add:
tgpu.resolve, kernels, or plugin output);tseynitround-trip would work and its tradeoffs (bundle size, startup cost, correctness);deepseek-v4.1-flash(free via Pullfrog for OSS) | 𝕏