mirror of
https://github.com/invoke-ai/InvokeAI.git
synced 2026-10-12 02:16:43 +00:00
Alexander Eichhorn
303e281778
- Every capability reader subscribes to the table instead of latching the fallback: the canvas grid, the dims sync, the topbar, regional guidance and gallery import. Reads happen inside the store selector -- React Compiler memoises a bare policy call in a component body for the component's life. - The Generate widget resolves again after a successful retry, and its loading and error states are announced and keep focus on the retry button. - Anima no longer offers FLUX VAEs: same channel count, different basis, and the decode is measured at 6.10 dB PSNR against 46.18 for the correct one. - The Qwen VAE branch dispatches on positive layout markers, loads in bfloat16 without a network round trip, and raises on a checkpoint that is neither. - webv2 reads VAE compatibility from the served table rather than a fourth hand-written copy, and joins the dimension grid on (base, variant).
43 lines
2.2 KiB
Plaintext
43 lines
2.2 KiB
Plaintext
# Auto normalizes line endings on commit so devs don't need to change local settings.
|
|
# Only affects text files and ignores other file types.
|
|
# For more info see: https://www.aleksandrhovhannisyan.com/blog/crlf-vs-lf-normalizing-line-endings-in-git/
|
|
* text=auto
|
|
docker/** text eol=lf
|
|
|
|
# The frontend formatters read the file on disk, so a CRLF working tree makes `format:check` fail
|
|
# for every file -- including ones nobody touched. That leaves contributors on Windows unable to run
|
|
# the same check CI runs, which is how formatting errors reach CI in the first place.
|
|
#
|
|
# Only the working copy changes: these files are already stored with LF (`* text=auto` normalizes on
|
|
# commit), so nothing in history is rewritten.
|
|
#
|
|
# It is not automatic on an existing checkout, and nothing warns you. Git applies `eol=lf` when it
|
|
# writes a file, and a pull only writes files whose content changed -- so a CRLF working tree
|
|
# becomes a mixed one: the files this revision touched are LF, the thousands it did not stay CRLF,
|
|
# and `format:check` keeps failing, now on files nobody edited. Rewrite the rest once, from the
|
|
# repository root, with nothing you want to keep uncommitted (this discards working-copy changes;
|
|
# `git stash` first if you have any):
|
|
#
|
|
# git rm --cached -r .
|
|
# git reset --hard
|
|
#
|
|
# `rm --cached` empties the index without touching disk; `reset --hard` then rewrites both from
|
|
# HEAD, which is the only step that applies these rules to files whose content did not change.
|
|
# `git status` is clean afterwards on a checkout that was already converted.
|
|
invokeai/frontend/**/*.ts text eol=lf
|
|
invokeai/frontend/**/*.tsx text eol=lf
|
|
invokeai/frontend/**/*.mts text eol=lf
|
|
invokeai/frontend/**/*.js text eol=lf
|
|
invokeai/frontend/**/*.jsx text eol=lf
|
|
invokeai/frontend/**/*.mjs text eol=lf
|
|
invokeai/frontend/**/*.cjs text eol=lf
|
|
invokeai/frontend/**/*.json text eol=lf
|
|
invokeai/frontend/**/*.webmanifest text eol=lf
|
|
invokeai/frontend/**/*.css text eol=lf
|
|
invokeai/frontend/**/*.html text eol=lf
|
|
invokeai/frontend/**/*.md text eol=lf
|
|
invokeai/frontend/**/*.yaml text eol=lf
|
|
invokeai/frontend/**/*.yml text eol=lf
|
|
tests/test_model_probe/stripped_models/** filter=lfs diff=lfs merge=lfs -text
|
|
tests/model_identification/stripped_models/** filter=lfs diff=lfs merge=lfs -text
|