github/vim-ale
1
0
Fork 0
mirror of https://github.com/webinstall/webi-installers.git synced 2026-04-06 10:26:49 +00:00
Code Issues Packages Projects Releases Wiki Activity

docs: update ANSWERS.md and GOER.md with FORMAT CHANGE fix summary (6,149→~3,200)

Browse Source
This commit is contained in:
AJ ONeal
2026-03-11 17:28:11 -06:00
parent 4f09649d30
commit 1dbcd312b8
2 changed files with 48 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

35
ANSWERS.md
View File

@@ -6,6 +6,41 @@
- [x] **Issue 4 — darwin-universal (Hugo)**: Go side correct (`universal2` in `CompatArches`). Node-side needs `universal2` in WATERFALL arch fallback.
- [x] **Issue 5 — Static musl → `libc='none'`**: Fixed in `classifypkg`. Rust `-unknown-linux-musl` → `none`. Hard-musl packages (node, bun, pwsh, julia, postgres) keep `musl`. See GOER.md for full verification.
## Response to QUESTIONS.md (2026-03-11, 17:27 update)
Committed `4f09649`. All ARM/MIPS/solaris/amd64-micro fixes applied. Cache copied.
**Root causes fixed (6,149 → ~3,200 FORMAT CHANGE warnings):**
1. **solaris/illumos/sunos (1,483 drops)**: classify.go was lumping all three to OSSunOS.
Node triplet.js treats them as distinct. Split into 3 separate OS patterns. ✓
2. **mipsle→mipsel (1,203 drops)**: legacyFieldBackport now translates. ✓
3. **mips64le→mips64el (1,050 drops)**: Same fix. ✓
4. **armhf→armv7 (152 drops)**: legacyARMArchFromFilename checked "gnueabihf" before
"armv7", so "armv7-unknown-linux-gnueabihf" got armhf instead of armv7.
Fixed: check "armv7" first. ✓
5. **armv6→armel (412 drops)**: Gitea "arm-5" filenames → patternToTerms → "armv5" → armel.
Go's `\barm\b` regex tagged them armv6. legacyARMArchFromFilename now detects "arm-5"→armel. ✓
6. **armv6→armv7 (90 drops)**: Gitea "arm-7" filenames. Same fix → "arm-7"→armv7. ✓
7. **armv6→armhf (14 drops)**: shellcheck "armv6hf". tpm['armv6hf']=ARMHF. Added "armv6hf"→armhf. ✓
8. **mips64r6 (12 drops)**: classify.go mips64 regex matched mips64r6 as substring.
Added ArchMIPS64R6/R6EL patterns before mips64 in both classify.go and buildmeta.go. ✓
9. **amd64_v2 (2 drops)**: classify.go ArchAMD64v2 regex had `amd64v2` not `amd64_v2`.
Changed to `amd64[_-]?v2/v3/v4`. ✓
**Remaining ~3,200 warnings are all E_MISSING_OS/E_MISSING_ARCH** (source tarballs,
iterm2 no-OS filenames, cmake IRIX/sunos-sparc64, git MinGit, dashcore) — cannot be
fixed without either excluding entries or adding per-package hardcodes for obscure platforms.
Cache copied to your `_cache/2026-03/` — please re-test.
## Response to QUESTIONS.md (2026-03-11, 16:46 update)
All issues fixed in commit `3756bd8`. Cache regeneration needed.

15
GOER.md
View File

@@ -63,5 +63,16 @@ Not backporting these — Go is correct, production has bugs.
- **atomicparsley**: `AtomicParsleyAlpine.zip` — needs per-package classifier config (hard musl).
RESEARCHER confirmed: needs a `releases.conf` with asset pattern overrides, not generic detection.
- **Hugo .pkg**: macOS v0.153+ only ships `.pkg`. Latent bug in both Go and production.
- **~7,400 informational warnings**: filename/arch mismatches (armhf vs armv6, etc.) from
the Node build-classifier re-parsing filenames. Don't block resolution; pre-existing limitation.
- **~3,200 remaining FORMAT CHANGE warnings** (down from 6,149): all E_MISSING_OS/E_MISSING_ARCH
from source tarballs (cmake, git), iterm2 no-OS filenames, dashcore, cmake obscure platforms.
Cannot reduce further without per-package hardcodes or exclusion rules.
## Classifier fixes (2026-03-11, 17:27)
- [x] solaris/illumos/sunos split into 3 separate OS patterns in classify.go
- [x] mipsle→mipsel, mips64le→mips64el in legacyFieldBackport
- [x] ARM priority: armv7 before gnueabihf in legacyARMArchFromFilename
- [x] Gitea arm-5→armel, arm-7→armv7 in legacyARMArchFromFilename
- [x] armv6hf→armhf in legacyARMArchFromFilename
- [x] ArchMIPS64R6/R6EL added to buildmeta.go + classify.go (before mips64)
- [x] amd64[_-]?v2/v3/v4 regex in classify.go (matches underscore form)