Chiel Douwes @[email protected]

I like optimizing stuff, and research about abstract data structures and algorithms.

[Chiel Douwes @[email protected] @chieltbest]

Due to recent events on twitter I am going to be moving to mastodon; please follow me @[email protected] for updates from me.

[Chiel Douwes @[email protected] @chieltbest]

A platform is a function of it's moderation, and I do not wish to participate in the platform that twitter is going to be just like I don't want to participate in sites like 4chan or kiwifarms. I'm sorry to those that disagree, but those are my reasons.

[Chiel Douwes @[email protected] @chieltbest]

Due to recent events on twitter I am going to be moving to mastodon; please follow me @[email protected] for updates from me.

[Chiel Douwes @[email protected] @chieltbest]

@ericniebler @pdimov2 @seanbax @odinthenerd About that... for metacheck I might actually want to create a list of all possible test cases which would probably involve some sort of cartesian product + transform + unique, which could easily be over 10k elems. That's essentially the reason I haven't done that so far.

[Chiel Douwes @[email protected] @chieltbest]

@odinthenerd This seems like a trick question to me

[emBO++ @emBOconference]

Things are getting even more fancy: @chieltbest about metachecks

[Chiel Douwes @[email protected] @chieltbest]

@odinthenerd @dascandy42 Either way it's four specializations, but one is a single memoized lookup and alias call more. Besides that, volatile is only ever used in tests anyway so combined is definitely faster.

[Chiel Douwes @[email protected] @chieltbest]

@Guriwesu @Sickeroni That's true, but if you don't inherit from the iterator you have to conditionally implement all the operations depending on the original iterator type using SFINAE. I'd wager that implementing a full version of the indexed wrapper would be several hundered lines of code that way.

[Chiel Douwes @[email protected] @chieltbest]

for (auto [i, v] : indexed{arr}) {
https://t.co/hP7PT2oeMY
So this works, except standard iterators are inconsistent, so it doesn't.
Props to @Sickeroni for the inspiration.

[Chiel Douwes @[email protected] @chieltbest]

@Manu343726 Probably because it wasn't very practically usable before this, it'd take about 8Gb of RAM to compile the complete kvasir::mpl test suite. Right now it's down to about .5Gb or so, not sure.

[Chiel Douwes @[email protected] @chieltbest]

Metacheck 1.1: now with multiple translation unit support, for multi core compilation and reduced memory usage. https://t.co/DErOBxa3uD 🎉🎉

[Carlos van Rooijen @carlosvrooijen]

Conference organizers beware! @chieltbest is now doing talks!

[Chiel Douwes @[email protected] @chieltbest]

Leaving @emBOconference at last, lots of interesting talks about resource constrained code and zero cost abstractions, i appreciate the clear vision on modern c++ design there was throughout all of the conference.

[Chiel Douwes @[email protected] @chieltbest]

Time for my first blog post, about the website internals and content: https://t.co/B2EqckKWLp. Posts about C++ will follow later on.

[Chiel Douwes @[email protected] @chieltbest]

I can now be reached at [email protected], hurray for domain names and postfix!

[Odin Holmes @odinthenerd]

I feeling quite addicted, is there a list of all C++ conferences and user groups?

[Chiel Douwes @[email protected] @chieltbest]

@TartanLlama Why not have the map function automatically switch based on the return value? I'd think no-one would want a nested optional anyway.

[Chiel Douwes @[email protected] @chieltbest]

@JasonRice_ @odinthenerd The documentation can be made by building the standardese cmake target, it might be a good idea to put those files online somewhere.

[Chiel Douwes @[email protected] @chieltbest]

@odinthenerd Is "metaprogramming" a single word yet? If so ::cml would also be an option.

[Chiel Douwes @[email protected] @chieltbest]

@EscJaeger @odinthenerd We haven't put out a release yet as we want to freeze the public interface first, so the master branch is still an early/concept version.