[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

qchateau updated this revision to Diff 374654. qchateau added a comment. - Rebase - Add cli argument to set the cache size - Reduce default cache size to 1 (to allow fast open/close/reopen) Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D97417/new/

[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

qchateau added a comment. Herald added a project: clang-tools-extra. Have you guys been giving some thoughts to that patch ? I've been using it in my daily work since I submitted the patch, and I'd not go back Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION

[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

qchateau updated this revision to Diff 327923. qchateau added a comment. fix bad arc diff Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D97417/new/ https://reviews.llvm.org/D97417 Files: clang-tools-extra/clangd/ParsedAST.cpp

[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

qchateau updated this revision to Diff 327922. qchateau added a comment. rebase on main, fix formatting Repository: rG LLVM Github Monorepo CHANGES SINCE LAST ACTION https://reviews.llvm.org/D97417/new/ https://reviews.llvm.org/D97417 Files: clang-tools-extra/clangd/TUScheduler.cpp

[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

qchateau added a comment. I am much more afraid of providing bad results than I am afraid of degrading performance. I mean, eventually the "real" preamble is built, and the results are just as correct as before, but it may yield incorrect results until we invalidate the AST. That is especially

[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

qchateau updated this revision to Diff 326816. qchateau added a comment. - [clangd] make more compile commands compatible - [clangd] ignore incompatible preamble I improved command line compatibility detection (faster & matches more files) and I had to blacklist some preambles. Namely, when

[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

kadircet added a comment. In D97417#2588101 , @qchateau wrote: > Well indeed do some extra work if we elect a compatible but almost useless > preamble. We'll basically do the work twice (but at least we do it > concurrently \o/). I am afraid this can

[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

qchateau added a comment. The hardcoded 5 premables can indeed be changed, I did not want to waste time on coding a configuration logic at such an early stage. Well indeed do some extra work if we elect a compatible but almost useless preamble. We'll basically do the work twice (but at least

[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

sammccall added a comment. @kadircet who was doing (or planning!) some measurements around this idea. Maybe it's possible to use this patch for them? My guess about risks: - currently, preamble patching covers relatively few extra headers in practice. Here we're likely to end up with larger

[PATCH] D97417: [clangd] use a compatible preamble for the first AST built

qchateau created this revision. qchateau added a reviewer: sammccall. Herald added subscribers: usaxena95, kadircet, arphaman, javed.absar. qchateau requested review of this revision. Herald added subscribers: cfe-commits, MaskRay, ilya-biryukov. Herald added a project: clang. Keep a store of the