Anonymous user
Talk:Distributed programming: Difference between revisions
→Insufficiently general?: ok, more to say
(→Insufficiently general?: also expanded task description) |
(→Insufficiently general?: ok, more to say) |
||
Line 12:
Hi. I wrote the original task. My intent was that it should show the use of ''distributed programming'' as opposed to simply being able to use the network — we have other tasks for that. The facilities used to accomplish the task should be suitable for ''performing a complex task distributed across several machines''. Or something like that. Trying to be sufficiently specific about sufficiently general, I'd say it should be a protocol/library which at least supports a reasonable set of common data structures, and preferably has a notion of messages/RPCs, such that the author of the example does not have to ''invent a protocol'', especially not a data serialization scheme, as opposed to just spreading their program across the network.
I've also expanded the task description a bit. I tried to avoid wording it exclusively.
As to your specific examples — I would say that anything where the application has to think about "data blocks" is far too low-level. Passing XML chunks around might qualify — ''if'' it goes with a library/facility for conveniently using XML as a data structure, ''and'' has some reasonable way to have communications between multiple independent components of the distributed program, i.e. remote object/subsystem identifiers.
I'm thinking that it might have been a good idea to define a specific application to be written distributedly, rather than leaving it as "just pass messages"; one which makes the task easiest to accomplish given the sort of facilities this task is intended to demonstrate. Maybe I should write a draft task "Distributed programming 2"?
—[[User:Kevin Reid|Kevin Reid]] 04:20, 18 June 2011 (UTC)
| |||