Jan de Muijnck-Hughes

@jfdm@discuss.systems
326 Followers
230 Following
2.5K Posts
Lecturer of type-driven approaches to trustworthy-systems (CyberSecurity) at Strathclyde. Professionally interested in PL & FM Methods; socially interested in coffee, politics, music, the outdoors, sci-fi, high fantasy, & much much more! My work doesn’t define me; it is not my identity.
wwwhttps://tyde.systems/
Pronoun’she/his
Locations🇳🇱 🏴󠁧󠁢󠁷󠁬󠁳󠁿 🇬🇧 🇪🇺
I also wish #Dafny did not pollute the source directory by default... please use a 'build' dir... you can emulate for executables but not for individual files in a project...

I am always struck by the strange quirks of #Dafny imperative and functional syntax:

Take the syntax for function/method prototypes:

```
method Foo(a : int, b : int) returns (res : int)
function foo(a : int, b : int) : int
function fooAlt(a : int, b : int) : (res : int)
```

and now anonymous functions:

```
method Foo<A>(f : (A) -> B, a : A) returns (res : B)
```

I wish Dafny would have some sugar that harmonises the use of `->` above in favour of `:`. Colon implies 'has type' not returns a type...

```
Foo(a : int, b : int) -> (res : int)
```

it can, however, compile and fine with Python...

TBH, this may be a Dafny issue, but it may be a #NixPkgs #NixOS #Nix

It could be that Dafny expects dotnet to be in a certain place, but then why do both versions behave differently...

Urgh #Dafny and #NixPkgs still not playing nicely...

Downloading direct from Github I get a working compiler.

Installation through NixPkgs means the resulting binary cannot find the local #dotnet install...

On my course, students like to say they know python.

I wonder how they feel about this:

https://gist.github.com/jfdm/fbb1f88065ab97b3c8e57ef322b1720e

*Four*

Not Quite Four but that is not the point...

Not Quite Four but that is not the point... GitHub Gist: instantly share code, notes, and snippets.

Gist

Urgh, our Office365 web interface has a shiny button for Copilot...

Can we just not...or at least have an option to hide it...

People who think their Agda code with no explanations is self-explanatory.

Oh... the image still from the video stops at an important slide!

Showing the Motivating Example Revisited...

Whilst I do not like tooting my own horn, if you were interested in my #TYPES2025 talk (Towards Being Positively Negative about Dependent Types) and could not make it.

Here you go:

https://www.youtube.com/watch?v=8xc50Lf6pXs

TYPES2025 - 4.7. Jan de Muijnck-Hughes - Towards Being Positively Negative about Dependent Types

YouTube

I hear ICFP is going to get FREX'd but with the meeting of two conferences, fools seldom differ!

Well done all.

×
@oblomov Let's do it the other way. AI as an optin instead of an optout.
@JauneBaguette that's actually an interesting proposal. Might be worth putting it in the GitHub issue CC opened for comments.

@oblomov > Github

Yeaaah… no. Do it if you want, I'm not going to that place ever again.

@JauneBaguette maybe contact them here then (but I don't know if they have anyone reading the comments at \creativecommons@mastodon.social