LDM Notes for May 12th, 2025

[333fred]

https://github.com/dotnet/csharplang/blob/main/meetings/2025/LDM-2025-05-12.md

Agenda

• Collection expression arguments
  • Empty argument lists
  • Constructor binding behavior
  • Syntax
• Dictionary expressions

[AlgorithmsAreCool]

I'm bummed that args didn't win over with. My intuition says that with has a connotation of a copy-constructor so it reads to me like we are copying from another collection, which is not the case. args is hard to interpret as anything but arguments which doesn't leave much room for ambiguity.

[AlgorithmsAreCool]

Respectfully, it sounds like your argument is "you might be confused a little, but you'll get over it", which doesn't feel like a strong counter.

[CyrusNajmabadi]

Respectfully, we looked at both. And subjectively felt that 'with' was better. We had people on the team who felt that 'args' was subjectively better. It was debated, and 'with' won out. There is no strong counter here, because the pros/cons for one versus the other are vanishingly thin.

args is hard to interpret as anything but arguments

It's easy to interpret as something different. Up until now, it's always meant the args passed into 'main' (and we even codified that as a well known keyword for top level statements). Now it means something different "a marker around arguments passed to something else". There were genuine concerns from people that they actively felt it was confusing as their mental view felt that this was a reference to those args.

Ultimately, we felt these were both fine options, and with narrowly won out. For a feature that data shows will be used 1:1000 (or less) in code, and often zero times in some codebases, we were not going to expend excess energy going in circles on this.

[AlgorithmsAreCool]

Yeah, it is definitely highly subjective and a minor quibble. And lord knows I've been dead wrong about LDM decisions before (*cough* NRT).

[jnm2]

you might be confused a little, but you'll get over it

Is this not the case for everyone encountering new syntax, for every new feature?

[AlgorithmsAreCool]

you might be confused a little, but you'll get over it

Is this not the case for everyone encountering new syntax, for every new feature?

yes, to greater and lesser degrees