Am Donnerstag, den 12.11.2015, 19:39 +0100 schrieb Virgile Prevosto:
> 2015-11-12 18:26 GMT+01:00 Daniel Bünzli <daniel.buenzli@erratique.ch>:
> > Le jeudi, 12 novembre 2015 à 16:37, Leo White a écrit :
> >> > This misses the point. By using `with` I can instruct the compiler that I don't want my program not to compile if I add or remove fields. In the current system I can do this with any update I want to make to a record except if for some strange reason that update happen to touch all the current fields of the record.
> >>
> >> I don't particularly wish to argue for or against using this warning, but I think it is worth noting
> >> that this is not really any different from:
> >
> > […]
> >
> > It's quite different: you have dead code in your example.
> >
> 
> But in the record case, the expr in { expr with ... } will indeed be
> evaluated (thus is not dead), but its result will never be used, which
> is barely better. In the (admittedly contrived) example below,
> investigating the cause of the warning would save you from a runtime
> exception:
> 
> type foo = { a : int; }
> let rec x = lazy { a = (Lazy.force x).a }
> let y = { (Lazy.force x) with a = 42 }
> 
> Now that I think of it, it is more like warning 26, with the exception
> that in the present case, the expression is usually a variable and
> thus already evaluated.

Maybe print the warning only in the non-variable case? And rephrase the
warning to "the expression before 'with' is not evaluated"?

Gerd

> 
> Best regards,
> -- 
> E tutto per oggi, a la prossima volta
> Virgile
> 

-- 
------------------------------------------------------------
Gerd Stolpmann, Darmstadt, Germany    gerd@gerd-stolpmann.de
My OCaml site:          http://www.camlcity.org
Contact details:        http://www.camlcity.org/contact.html
Company homepage:       http://www.gerd-stolpmann.de
------------------------------------------------------------