Talk:Pipeline programming

Can pipeline programming be considered a subset of dataflow programming, in which modules are limited to only one input and one output? --Pezezin 22:59, 5 January 2007 (UTC)[reply]

Arguably, pipeline programming is really a limited form of function-level programming. Note that functional programming is NOT function-level programming. In essence, the program is really a pipeline of programs 202.163.215.37 15:57, 22 May 2007 (UTC)[reply]

I don't think so. In general, pipeline programming is very much "value level" (in Backus's terms) or "dataflow", in that successive pipeline stages operate on values produced by the previous stages. There are some examples of hybrid systems, such as "Hartmann pipelines", that can reconfigure themselves on the fly, and thus show some function-level capability, but they still tend to be value-biased. Certainly the standard Unix pipeline paradigm lacks any kind of "conditional", with the possible exception of composition. RossPatterson 23:21, 22 May 2007 (UTC)[reply]