The goal is 8 bytes, which has a nonzero risk of collisions with huge indexes.
This patch should shake out any issues with truncation at all, we can lower
further later.
Details
Details
Diff Detail
Diff Detail
- Repository
- rL LLVM
Event Timeline
Comment Actions
We should watch out for the downstream backward compatibility.
| clangd/index/Index.h | ||
|---|---|---|
| 92 ↗ | (On Diff #170684) | maybe SHA1(USR) truncated at RawSize bytes? | 
| clangd/index/Serialization.cpp | ||
| 303 ↗ | (On Diff #170684) | A reminder from yourself: bump the binary version ;) | 
| 303 ↗ | (On Diff #170684) | maybe use RawSize here instead of hardcoding? | 
| clangd/index/Serialization.cpp | ||
|---|---|---|
| 303 ↗ | (On Diff #170684) | I think it's actually more important that this is concrete than that it's correct :-) |