👾

I develop. He/him.

[👾 @ashmind]

New places (no content yet):
• https://t.co/5jGSUbhrnv
• https://t.co/AeXjHUU1kk

[👾 @ashmind]

@ereliuer_eteer I would say, don't focus your best skills on the goal of breaking it (just in case). But also don't be cautious with anything you try as part of normal usage, SL is expected to handle it.

[👾 @ashmind]

@ereliuer_eteer There are some built-in protections against type explosions btw: https://t.co/qgqg3uPg70

If you find something that still causes a problem, please let me know and I'll extend those further, the system is designed to be resilient against user input.

[👾 @ashmind]

@ereliuer_eteer Thanks for reporting! However I don't think it's caused by your code — there is a nightly build and it seems latest main didn't build correctly, so likely just a timing coincidence.

[👾 @ashmind]

@terrajobst I wonder where the tech people will go — keen to follow MS .NET ppl on whatever is the new service (hopefully not LinkedIn though).

[👾 @ashmind]

@davidfowl Fair point.

Even if it is an explicit opt-in, I would still pick global opt-in over per-endpoint config.

But the question of opt-in vs opt-out and the startup cost is very interesting design problem, I'll think about it.

[👾 @ashmind]

@davidfowl Yep looks like the best option, tbh could be a sensible framework default, without any call? (though could be a breaking change, not up to date on latest ASP .NET).

Allowing individual endpoint to process its validation errors in a custom way feels like an edge case.

[👾 @ashmind]

@PeledZohar @jaredpar @terrajobst We used it historically before LINQ expression trees existed. E.g. Where(Client.Age > 30) where > generated some kind of BooleanQueryExpression which was then converted into SQL.

Not sure if there is much value in modern uses though.

[👾 @ashmind]

@KirillOsenkov No need for Roslyn fork, can just use IgnoresAccessChecksToGenerator:
https://t.co/pyN4iQuEUS

[👾 @ashmind]

@chriscoyier I might be a bit biased, being originally from Russia. But people do not choose what country to be born to. If the author of a library or technology personally chooses to promote heinous views, that can be a reason to drop the tech. Their birth country, not so much.

[👾 @ashmind]

@davidfowl Great question, I had to think about it.
I think still not OK, due to encapsulation.

public IX Dependency { get; init; } — get breaks encapsulation

public IX Dependency { private get; init; } — technically doesn't, but very weird/awkward pattern

[👾 @ashmind]

This works for method visits as well, e.g. here is a method that gets called two times. You can see what called it that time, and what return value was produced for that specific call.

[👾 @ashmind]

Recently I was thinking about "time-travel" alternatives to traditional debugging for SharpLab's continuous Run.

Here's one outcome. SharpLab used to dump all loop values at once, but that wasn't too useful. Now you can select a specific iteration and see values and flow within.

[👾 @ashmind]

@sonergonul Thanks for raising! This is now resolved in newer branches (including main).

[👾 @ashmind]

@davidfowl Makes sense — e.g. we have an app where migrating WebForms part is multiple years with no customer value.

I've considered same, so we could reap immediate tech benefits (C# 10, Mac dev, Linux host), and leave fig rewrite (also simpler, no YARP) to future UX-oriented initiatives.

[👾 @ashmind]

@davidfowl Yep hoping it evolves, last proposal I saw were sizes only:
https://t.co/nqAS2tRXhk

[👾 @ashmind]

@benblo42 Thanks! It is definitely on my mind, but at the moment — not yet (and likely not this year, but we'll see). Though I would be open to people contributing it.

[👾 @ashmind]

@OneWholesomeDev Yes, eventually, however it's not in my immediate backlog (need to do some maintenance and fixes first).