>> Run Pkg.update() in my Julia terminal. >> It automatically

Julia's package manager is crazy. If I install a package with a dependency on python packages, it will download and install anaconda by itself and conda install the python packages you need as a dependency.

So disappoint.

File: 7bc360_5672012.jpg (279KB, 2241x1500px) Image search: [Google]

279KB, 2241x1500px

>>59331026
Neat and full circle, is the build system also a library?

>>59331026
>system-wide or even user-wide programming language package management instead of project-wide for best granularity and least side-effects
That's why DUB is the best programming language package manager.

>>59336051
So basically, what everything else does.

Now DUB only needs a good programming language with good reference compilers.

>>59336124
DMD is great though, anon.

>>59336574
I don't think so. Last time I checked it was a crashy, not-really optimising piece of shit that relied on DMC and on some promises from Symantec.

>>59336666
Didn't DMD switch purely to D code a while back?

>>59336971
Yeah, "last time I checked" obviously means 8 years ago for that guy, but apparently he thinks bashing D on online Tibetan Titanium-welding imageboards without rechecking it every at least couple of years is a reasonable thing to do.

>>59336971
I wouldn't know, I haven't been following its development for a year or so.

But even in this case, it wouldn't be able to compete with LLVM or GCC as backend. That's not something I'd want as a reference compiler for a language that now aims at being a C++ replacement.

The greater overall problem seems that the whole thing seems to be the ego trip of Bright while aimlessly fucking around for the last decade or so.
On the other hand, C++ commitee doesn't have their shit together, so whatever.

>>59337129
I don't really see anything wrong with the reference compiler not optimizing very well. It compiles super quick for unit tests, and LDC works with dub as well.

>>59337081
>Yeah, "last time I checked" obviously means 8 years ago for that guy, but apparently he thinks bashing D on online Tibetan Titanium-welding imageboards without rechecking it every at least couple of years is a reasonable thing to do.

Pretty sure that was just the frontend of DMD while the backend got replaced not quite such a while ago.
Actually nevermind, I remembered it wrong, DMDs garbage distribution for Windows just installs DMC.

>on online Tibetan Titanium-welding imageboards
kek

BTW: Is it still true that quite a lot of libraries can only be compiled with DMD?

File: plb-lang.png (51KB, 652x383px) Image search: [Google]

51KB, 652x383px

>>59337193
Even without optmizing a lot, it's still faster than C++ and sometimes even C in many use cases.

>>59337214
I regularly use dlangui, vibe.d, dpq2, and a shit ton other libraries. I prefer LDC for compiler. Never had troubles compiling anything.

>>59337193
I think the concept of reference implementations is garbage for anything but the most trivial standards. It barely works for codecs.

>>59337218
We were talking about DMD not optimizing very well ya git, not LDC or GDC.

>>59337246
Who gives two shits about DMD? It's proprietary crap!

>>59337258
Only because Bright isn't the sole copyright owner.

>>59337218
And by that you mean cherry-picked benchmarks and stuff that would require too painful metaprogramming in C++.

>>59337285
I actually wonder how Symantec could become half the copyright owner.

Did he secretly code it in his worktime and got busted?

>>59337300
Well D's templates being a few orders of magnitude less painful than C++'s is one of its strong points.

>>59337334
Pretty much.

>>59337354
wtf

>>59336666
>>59337246
>>59337300
It's not designed for that, you dumbasses. It's made for QUICK compilation, which is what you want when you're developing big projects: you don't want to wait the entire morning to build the whole thing just so you can test it. Developer time is costly.

GCC does NOT offer you that because it focuses so hard on being the Jeremy Clarkson of the open source community (MUH SPEEEEEEEED).

>>59337468
So I take it incremental compilation isn't possible?

>>59337218
>faster than C++ and sometimes even C
Don't joke like that, anon.

>>59339944
This, obviously C++ is at worst same as C.