this "when to use section" is misleading

[hiredman]

I think a when to use section would be nice to have, but the current one only makes sense as a set of restrictions applied to fold, not reducers in generally.

If you look at http://clojure.com/blog/2012/05/08/reducers-a-library-and-model-for-collection-processing.html under the heading "More Opportunity (i.e. Work)" rhickey explicitly called out reducers as a mechanism for solving "the lazily dangling open resource problem"

[puredanger]

Re fold, that makes sense, but we could just change the heading rather than removing it.

Re the other point, are you suggesting that this should be added? Trying to square that comment with the suggested change which doesn't seem to address that.

[hiredman]

I am pointing to the blog post about reducers as, uh, supporting testimony, that the advice in the "When to use" section is not valid for reducers in general. I am not suggesting that needs to be added. Alternatively the "When to use section" could be re-titled as "When to use fold".

I have been advocating reducible collections as a great way to model resources but today someone in slack pointed out this "When to use section" as seeming to be counter to that. Given my understanding of reducible collections, and the initial blog post announcing them, I think the "When to use" section is in error. Its advice about avoiding IO and blocking make sense for fold but not for reduce in general. Given how much nicer I think reducible collections are for modeling resources, and how hard it is the pry seqs out of the hands of Clojure programmers, it would nice if the Clojure website didn't give (what is in my opinion) incorrect advice on the topic.

[puredanger]

Made some alternative mods in c1ef589