Every now and then I want a set type, but I have to either use my
own library for it, or use a commonly known associative array
trick. I really think (unordered hash) sets should be part of the
standard library. While I was working on something else, I
thought of one way of adding them to the s
The signatures only had 'private' in there because I copy and
pasted them out of something I was working on. (Because they
work.)
On Sunday, 29 March 2015 at 09:17:17 UTC, w0rp wrote:
3. I'm using 2.066 still, and some things like byKey aren't
@safe and so on. These things need to have attributes set. (As
always, opApply probably lacks one of @nogc, @safe, pure,
nothrow because of the delegate types.)
I also just realis
w0rp:
Every now and then I want a set type,
I think you can implement good enough sets in Phobos, do you
agree?
But you can't do the same with tuples. So I prefer sets in Phobos
and tuples in the language.
Bye,
bearophile
On Sunday, 29 March 2015 at 14:23:10 UTC, bearophile wrote:
w0rp:
Every now and then I want a set type,
I think you can implement good enough sets in Phobos, do you
agree?
But you can't do the same with tuples. So I prefer sets in
Phobos and tuples in the language.
Bye,
bearophile
+1
On Sunday, 29 March 2015 at 19:32:48 UTC, weaselcat wrote:
Is std.container in need of more containers?
These
https://github.com/economicmodeling/containers
have Phobos-compatible license (Boost) and, with some work, I
guess some of them could be merge into std.container, in your
case *hash
On Sunday, 29 March 2015 at 19:39:36 UTC, Nordlöw wrote:
On Sunday, 29 March 2015 at 19:32:48 UTC, weaselcat wrote:
Is std.container in need of more containers?
These
https://github.com/economicmodeling/containers
have Phobos-compatible license (Boost) and, with some work, I
guess some of t