Tailored Swift – coming soon to a cloud near you

Jonathan "Jono" Guthrie, senior software engineer for PerfectlySoft, dropped by The Register's office in San Francisco to discuss the merits of the Swift programming language. The Apple-created Swift has attracted significant attention from developers since the iPhone maker released it as an open source project last year, but …

[P. Lee]

>So people had to have virtual machines that scaled vertically and then all of a sudden they have to scale horizontally and it becomes very expensive very, very quickly.

I have to question this. If you're using something like (the very cool but expensive) F5 then yes, its expensive to set up. We really need some cheaper horizontal scaling. If you're running public cloud, renting your servers, then it is expensive. If you are licensing per CPU or core, then it is expensive.

However, If you run your own IT, if you have some cheapish load balancing and you're running open-source, then CPU and memory are not that expensive to scale out. Cloud is based on over-subscription - constant high usage ruins that model and providers cost things accordingly. Commercial software licensing is also often based not on utility to the customer, but technical characteristics. That encourages vertical scaling and discourages horizontal scaling. Just like the cloud providers, if you are scaling out, then acquire the technical skills and use lots of the relatively cheap tin.

[Alphabet Soup 1]

Namespace error

Clicked on this to read about OpenStack's widely-used scalable object storage component 'Swift" - does the cloud have room for two Swifts?

[JLV]

>So with Node, with Python, with all of these other sorts of languages, with Go, you just go to GitHub and grab some code or import it as a dependency.

>Is there enough data yet to make language choices based on performance?

Thinking out loud... Python, being a slow-a** language, has always been very, very, keen on shelling out to something faster if at all possible. You have tons of bindings to C and other stuff. Plus, you can actually write fairly fast Python, by using language builtins that are uber-optimized in... C. Think Dictionaries for example. Except for the builtins, this is all fairly transparent to the core language (I think - haven't used much of them myself), these are add-ons facilitating integrations, not core features.

So, Python has had to these trade-offs of productivity -> performance. At the extreme end, you shunt off your really heavy bottlenecks in C and keep Python where you can.

Could Swift programs, at least initially, go the reverse way and shell out development-intensive, but not particularly time sensitive bits - or just where there are existing libraries elsewhere - to say, Python? Or something else. With automated asset grabs / package managers sucking up the dependencies from the language's package managers?

It would be an odd approach to take - but after all machine perf vs developer time tradeoffs don't always go in one direction.

[GeezaGaz]

Performance

>Is there enough data yet to make language choices based on performance?

Based on what was written in this article hardly. I mean Twitter didn't realise they had a problem with Rails until there were generating gazillions of messages every second.

Its not battle tested (yet) with a massive throughput site so it remains to be seen if its a *better* option.