I like it that they are adding more and more functionalities from J, probably because Roger Hui is involved in this. The namespace system in Dyalog APL works far better than J's locale, IMHO.
The only problem is that you can't get the source code. Look around, in this 21st century, how many interpreters are still closed sourced? I really worry about their long term survivability.
Closed source is less of a problem for me than the draconian licensing. Paying such steep costs to use in production is a no go for me. I could afford the $1500 personal productivity tool fee for just myself, but then none of my coworkers can use anything I write which probably doesn't help advertising. I really like the symbols more than J, but J ends up being more practical for me. I think if array langs are to take off, kx systems would have to make their software free and just charge for support, but I doubt they'll do that with how much money they're making.
I still don't get the companies that cannot take the lessons of Smalltalk vendors. It's no accident Borland could sell product and Smalltalk vendors died off with developer and deployment fees.
I still think there is a place for paid tools, but more JetBrain / old Borland then ParcPlace.
I've only used it enough to see that majority of the code I write in Dyalog APL will run there as well. Perhaps I should start in GNU APL in the first place, but Dyalog's environment is just nicer
I really like Dyalog. I paid for the personal license before they made it free, but then when I thought about developing things for fun, and possibly commercial reasons, the licensing put me off. I still play with Dyalog, but J is the better fit for my needs right now. I am glad Roger Hui is on board, and I hope it does the same thing J did, and open source the code.
I have a fondness for the characters in APL for no real objective reason. Maybe my love of symbols to make things concise and in one eyeful.
I skimmed the Wikipedia article on AVX, but have no clue what it means. Care to go into a little more detail? Targeting the additional instructions will slow things down you think?
AVX are vector instructions. They work on wide register (256bit) where you can stuff 8 32bit floats (or in later versions ints) or 4 (double, or longs not sure if AVX2 or later has support for that), or 16 chars (AVX2 iirc).
APL, as an array language, should, in theory at least, especially profit from automatic vectorization by the compiler. If the auto-vectorization works well, you get get a 4-16x speed-up for free (if you don't have data dependencies).
That seems to be about right. I really wish there was a decently maintained, open source, apl derivative with the source on GitHub. The nearest I could find was Kona but it looks like that has pretty minimal activity. It seems like anyone with the ability to make this a reality just creates a new variants , closes the source, and charges for the tooling (like Kerf).
The only problem is that you can't get the source code. Look around, in this 21st century, how many interpreters are still closed sourced? I really worry about their long term survivability.