Tomas Grosup
Tomas Grosup
@TheAngryByrd : I would love to have this return-type subsumption supported built-in, but I think that would just lead to fully admitting the runtime async the compiler. (i.e. this conversion...
Yes š .
@copilot : Make a new set of experiments, compare the time+memory results for: 3000 files with 3000 modules 1000 files with 3000 modules (i.e. 3 modules per file) 3 files...
@copilot : Use the compiler's local Activity module to add instrumentation about timing, contents and size of the data handled within CombineModuleOrNamespaceTypes . Use the 3000 files, 3000 modules, but...
@copilot : Create a dedicated Benchmarkdotnet benchmark for QueueList, simulate the scenario in question (i.e. 5.000 additions of a single element, one by one). There are benchmarks in this repo,...
@copilot : There are two problems with AllEntitiesByLogicalMangledName : 1. It is not cached with a mutable optional NameMap 2. The constructor for ModuleOrNamespaceType always builds EVERYTHING (see all the...
@copilot : "expected improvement" is a joke. Just do the measurement for real.
@dawedawe : Do you have an idea of what my going wrong here ?
@copilot : List all functions available for `List` in Fsharp.Core which are not yet covered by time+space complexity comments after your change. Did you miss any? What about members which...
@copilot : Execute what you have identified: After reviewing the codebase, I identified the following functions that are still missing time and space complexity documentation: List module functions without complexity...