github-mirror/Telegram-iOS
1
0
Fork 0
mirror of https://github.com/TelegramMessenger/Telegram-iOS.git synced 2026-07-05 19:28:46 +02:00
Code Issues Projects Releases Packages Wiki Activity Actions

CLAUDE.md: add public-Postbox-type inventory lesson to wave-selection guidance

Browse source 
Surfaced by wave 16 -- grepping only for Postbox/Network tokens undercounts
the hidden public-surface leaks (e.g., EngineMessageHistoryThread, PeerStoryStats).
Document the allowlist-grep recipe and the three paths for handling hits.

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
This commit is contained in:
isaac 2026-04-20 21:39:15 +02:00
parent 8e7014a7ed
commit 8215c057cf
1 changed files with 14 additions and 0 deletions
Download patch file Download diff file Expand all files Collapse all files

14
CLAUDE.md
View file

@ -102,6 +102,20 @@ Before selecting a wave's module list, grep each candidate for:
Tally impact from a sweep: dozens of consumer modules can become Postbox-free in a single commit. First run (wave 6): 782 candidates, 18 iterations, 183 survivors, **189 modules** newly Postbox-free. Re-run after every 2-3 facade-migration waves.
**Public-Postbox-type inventory (wave-11-pattern planning).** For wave-11-shape candidates (modules whose public init takes `postbox: Postbox, network: Network` purely for avatar/setPeer forwarding), grepping only for `Postbox`/`Network` tokens **undercounts** — public-surface types defined in Postbox can leak without ever naming "Postbox" literally. Wave 16 hit this: the plan missed `EngineMessageHistoryThread.Info?` and `PeerStoryStats?`, both Postbox-defined types whose names don't include "Postbox". Mitigation: build a Postbox-defined-public-types allowlist once, then grep the candidate module against it.
```bash
# Build allowlist once (or re-run if Postbox sources change):
grep -rhE "^public\s+(class|struct|enum|protocol|typealias)\s+\w+" submodules/Postbox/Sources/ \
| awk '{print $3}' | sed 's/[(:<].*//' | sort -u > /tmp/postbox-public-types.txt
# Then, for each candidate module, grep its sources for any of those names:
grep -rhoE "\b($(cat /tmp/postbox-public-types.txt | tr '\n' '|' | sed 's/|$//'))\b" \
submodules/<CandidateModule>/Sources/ | sort -u
```
Any hit in a public-surface position (field type, init param type, enum payload type, generic arg) that isn't already a documented typealias is a blocker. "Engine"-prefixed types can still be Postbox-defined — don't trust naming conventions, grep for the defining module. If the module hits only `Postbox` itself (i.e., literal `Postbox`/`Network` pair), it's a clean wave-11 candidate. Otherwise, decide per leak: (a) move the type to TelegramCore if it's a namespace-only class (wave 16a pattern — prototype: `EngineMessageHistoryThread`), (b) accept that the module can't become Postbox-free and ship a partial `engine:`/`stateManager:` collapse that keeps `import Postbox` (wave 16b pattern — `PeerStoryStats` is too baked into Postbox views to move cleanly), (c) abandon the candidate.
### Wave 1 outcome (2026-04-16)
4 modules done: `ChatInterfaceState`, `ChatSendMessageActionUI`, `ContactListUI`, `DrawingUI`.