January/February 2018 issue of acmqueue

The January/February issue of acmqueue is out now


  Download PDF version of this article PDF

ITEM not available


Originally published in Queue vol. 10, no. 7
see this item in the ACM Digital Library



Graham Cormode - Data Sketching
The approximate approach is often faster and more efficient.

Heinrich Hartmann - Statistics for Engineers
Applying statistical techniques to operations data

Pat Helland - Immutability Changes Everything
We need it, we can afford it, and the time is now.

R. V. Guha, Dan Brickley, Steve MacBeth - Schema.org: Evolution of Structured Data on the Web
Big data makes common schemas even more necessary.


(newest first)

Headinthebox | Sat, 24 May 2014 04:05:34 UTC

Ruediger, you are spot on. If you carefully read the section on distributed collections, you see that I do propose, and use in the code examples, that the APIs are asynchronous.

Ruediger Moller | Fri, 23 May 2014 10:44:05 UTC

What's completely overseen in the vision of transparent distribution of collections is the effect of latency. You won't be able to distribute synchronous interfaces effectively, as e.g. on a dictionary you'll get a complete network roundtrip latency added to your call.

What's needed on-top is a transition in programming and interface patterns from synchronous to asynchronous ('reactive') calls. And I'd say most languages and tools are poorly equipped to support asynchronous programming patterns in a convenient way.

Barfo Rama | Wed, 22 Aug 2012 22:36:55 UTC

I notice that html syntax is not allowed and will be removed. I suppose that implicitly comments on how exposing tracing and debugging to cloud interfaces will let random cyber-hoodlums iteratively redirect your credit card object containment to their Cayman Island banking monad while displaying little circles going around and around.

Ironically, the hoodlum behavior is more predictable than the objects.

David Broderick | Wed, 22 Aug 2012 13:31:44 UTC

I hate to argue against someone I admire as much as Erik, but I disagree. His argument makes sense only for 'mandatory' context and rules, and that's a minority subset.

If I'm querying for the length of something, is that 18mm or 18au? Units are mandatory context for something quantitative. But the moment the context or rules become situational (which ends up being a large amount of the time), the problem of using objects in this space means forcing context and rules which don't apply. And then you have something either useless or less effective.

The future, instead, is to model the context and rules as you do the data. Users can then simply query how much they want/need of the former.

Mark W. Farnham | Thu, 16 Aug 2012 20:53:59 UTC

"In the big open world of the cloud, highly available distributed objects will rule." Wow. What a leading statement.

For starters, you presume a world in which the cloud is highly available. For most infrastructure stacks of which I am aware, global network access is the weakest link regarding uptime, latency, and bandwidth.

Some of what you write after that is probably relevant, though the over verbose camel hump strings you posit tossing back and forth in all directions over the network to your global cloud will contribute to its failure.

As for generalizing Codd, the relational model is in fact a general model. Some folks allow particular (and so far in every case partial) implementations of the relational model to pollute understanding of the relational model. But Chris Date has already represented that argument better than I can.

Leave this field empty

Post a Comment:

© 2018 ACM, Inc. All Rights Reserved.