Proposal: Add (&) to Data.Function

[Heinrich Apfelmus]

Johan Tibell wrote:
> On Thursday, November 22, 2012, Joachim Breitner wrote:
> 
>> Hi,
>>
>> -1 for adding such an operator at all. Languages benefit from consistnet
>> idiom; too many ways of doing the same thing are not always good.
> 
> 
> -1 for me for the same reason. After looking at examples in this thread I
> don't find that the loss in readability is worth it.
> 
> Also for beginners I don't want to explain yet another way to write
> function application.

-1 from me as well.

The (#) operator is quite natural and useful in the  diagrams  library, 
but the thing is that even though it is *implemented* as function 
application, it is *semantically* not a function application. It only 
serves to furnish shapes and diagrams with additional properties; it's 
type is very restricted, so to speak.

Since the use cases mentioned (diagrams, lenses) are very similar, 
perhaps there is a general combinator (&) that does the job for both, 
but which has a much more restricted type and is not equivalent to 
reverse function application. Maybe an abstraction like "Settable 
functors", "Thingomorphisms with properties" or something like that.


Best regards,
Heinrich Apfelmus

--
http://apfelmus.nfshost.com