Dyalog ’17: Day 2 (Monday 11 September)

by Vibeke Ulmann

Focus on Dyalog APL the language – Monday 11th September 2017

Where Sunday is traditionally filled with workshops and hands-on experiences – the first proper day of the annual user meeting is Monday – and this year was no different.

CEO, Gitte Christensen opened the meeting and emphasised a few of the major new things that have been achieved since last year, namely:

  • RIDE Version 4
  • Embedded HMTL Rendering Engine and – as always –
  • Performance enhancements.

She highlighted the fact that Version 16.0 was the beginning of a tool chain for developing distributed applications – including Cloud computing.

The licence for the SyncFusion library has been renewed for another 5 years. So, for those working with SyncFusion, you will have the usual widgets for dashboarding and graphing to hand.

A new multi-platform developer licence is now available allowing for development on all platforms; Windows, Linux, Mac, and soon, Android.

The Tools Group has been expanded, and they are producing more and more examples and templates.

Dyalog is now producing live content (outside of the user meeting) in the form of webcasts – currently one a month – and Podcasts are also planned.

CXO Morten Kromberg gave us a look at the next generation APL.

Authors note: if you are wondering what the X in CXO stands for it is ‘Experience’.

Morten established that there is a general ‘climate change’ in the world of computing especially as cloud computing is now ‘THERE’. This means that performance once again becomes key, as the true cost of cloud computing is measured in Watts – meaning CPU and memory consumption. So, if you can reduce the footprint of your application, you can reduce the costs of cloud hosting. Another point made is that Cloud computing generally means Linux – as it uses less memory and, therefore, fewer Watts. Whereas macOS and Androids can be considered UNIXes.

Morten focused on the demand for a new generation of APL developers and more to the point, managers who are comfortable using APL, and APL programmers. There are a number of criteria that both groups need; the developer need a modern set of libraries to build upon and to be able to find them easily for example using Git. Whereas managers need test driven development, source code management, and continuous development cycles.

Not everyone is a familiar with Git, and some are even a tad intimidated by it. But there is much good to be said for Git. You can have a private area, as well as a public shared area. Dyalog currently has 25 public repositories on Git. More will follow over time.

CTO Jay Foad proceeded to outline how we can make some of the Dyalog dreams for the future come true.

Version 16.0 of Dyalog APL was released in June this year, and work on version 17.0 continues apace. Speculatively Jay highlighted some of the key areas in version 17.0 to be: scripting, language, performance, and bridges to Python, Julia, MATLAB and Haskell. More work on RIDE, GPUs (and Xion Phi), portability and Android, the Cloud, shuffle testing and PQA.

The Key Note speech before Lunch was presented by Aaron Hsu from Indiana University (USA). Aaron went through how you can escape the beginner’s plateau when starting to work with APL.

The key takeaways for yours truly was that Aaron has observed two attitudes to APL

1)     Never in my life

2)     Can’t imagine life without it

He also observed that there seems to be a ‘learning wall’ which we need to find a way to overcome.

A directory of best practices can give insight into why computer scientists, or those trained in traditional programming methods, often find APL jarring and difficult, whereas those with no prior training fall in love with APL and take to it like ducks to water.

Watch his presentation in which he walks through 8 patterns, he considers to be key for newbie APL programmers. We will announce when the presentation is available on Dyalog’s YouTube Channel later in the autumn.

After lunch, most of the afternoon was dedicated to looking deeper into some of the new feature/functionality and topics many APL programmers find of particular interest.

In the interest of enticing you to watch the presentations online when they’re posted on Dyalog’s YouTube Channel, this blog only touches the basics on a couple of the presentations.

John Scholes went through re-coding from procedural to denotative style and showed us how pure functions opens for code reduction when implemented. ‘Massaging’ the code was the new expression I came away with.

Roger Hui showed us how he has now managed to solve a 20-year-old problem: ‘Index-of’ on Multiple Floats. After having initially established – to much hilarity – that the best way of solving the problem at hand was to not introduce it in the first place in your code, Roger proceeded to show how it can now be solved. Intentionally I am not giving away what Rogers 20-year-old ‘Problem’ was. Let me just briefly mention: it has to do with X and Y.

The afternoon was rounded off with a user presentation by Kostas Blekos from the University of Patras (Greece), where a group of physicists have used APL for the research they did for a paper.

His initial premise was that Physicists + Programming = Disaster. On the other hand, physicists need to do a lot of programming, so when they were developing the basis for the paper, they wanted to find a (new) language that made it easier to do better (and faster) prototyping.

Lots of Kostas’ and his colleagues’ previous work had been done in FORTRAN and, as he said, we needed something a bit easier to work with and the choice was Dyalog APL. Outside of the ability to do fast prototyping, the terseness of the language was attractive, as were the close mathematical relations, which made it easy to understand.

What they learned was that APL is GREAT, suitable for fast prototyping, and for avoiding making mistakes. The quote of the day surely must be

In FORTRAN I could spend a whole day trying to find a missing comma……..

Asked if there were any downsides to APL, Kostas said no, not really, except it is difficult to convince people to use it.

Comments are closed.