WIP: Ensure we always cache diag results. #73199
Conversation
dotnet-issue-labeler
bot
added
Area-IDE
untriaged
|
|
||
| // Now, save all that data to the in-memory storage, so it can be retrieved later when making requests for | ||
| // other documents within the same project. | ||
| await SaveAllStatesToInMemoryStorageAsync().ConfigureAwait(false); |
There was a problem hiding this comment.
@mavasani i was hoping this would help. By making it so that we always save for any caller of GetProjectAnalysisDataAsync (currently there are two of htem).
However, it looks like this doesn't help because the code that attempts to load from teh cache seems to bail out for any number of reasons i can't make heads or tails of.
| // this is perf optimization. we cache these result since we know the result. (no diagnostics) | ||
| var activeAnalyzers = stateSets | ||
| .Select(s => s.Analyzer) | ||
| .Where(a => !a.IsOpenFileOnly(ideOptions.CleanupOptions?.SimplifierOptions)); |
There was a problem hiding this comment.
@mavasani this part seems very weird to me. We would get the statesets... but then extract out the analyzers that we'd actually want to run. I tried to change this to just filter down teh statesets in the first place, makign it so that we have a 1:1 mapping from stateset to the analyzers we run, and so we wouldn't have a case where we were looking at statesets taht were actually running/computing/caching data because we filtered out their analyzer.
| diagnostics.AddRange(analyzerResult.GetAllDiagnostics()); | ||
| await state.SaveToInMemoryStorageAsync(project, analyzerResult).ConfigureAwait(false); |
There was a problem hiding this comment.
this was removed because it happens within the call to GetProjectAnalysisDataAsync above now.
|
Closing out. we went with a project-level cache at the LSP pull diags level. |
Followup to #73198