Re: Curry module system

From: Wolfgang Jeltsch <>
Date: Fri, 17 Feb 2006 17:23:10 +0100

Am Freitag, 17. Februar 2006 11:16 schrieb Michael Hanus:
> Wolfgang Jeltsch wrote:
> > Note that the Haskell community repeatedly said that the current Haskell
> > record system is broken and this very community is searching for a better
> > alternative. Maybe we shouldn't include something into Curry which is
> > "broken".
> Thanks for the feedback. Could you comment a bit more a provide
> a link to explain why the record system is broken?

Not very much at the moment, unfortunately. I'm far away from being an expert
on this topic. I just noticed from reading mails on several Haskell mailing
lists that many people seem to consider Haskell's current record system a
mess, and picked up some of their arguments. I will ask on the Haskell
Mailing List if there is a web page or so which describes the problems with
the current system in a compact form. Currently, I don't know of such a

> Do you have the uniqueness of selector names in mind?

This is one of the problems. Others are the non-extensibility of records and
the fact that labels aren't first-class, if I remember correctly.

> [...]

> > > One other mini-proposal: I like to include the definition
> > >
> > > x_ = let x free in x
> >
> > But because it's so obscure, it shouldn't be used, in my opinion.
> I mean the name is obscure for meaningful functions
> or pattern variables so that it is not used in existing programs.
> However, the name x_ is reasonable to denote an anonymous free
> variable. Or any better suggestion for a compact name?

I would never name a global variable x or something similar. In fact, I also
don't name any local variables x. I'm very much in favor of descriptive
names like "free" (which is a reserved word, unfortunately).

> Best regards,
> Michael

Best regards,

curry mailing list
Received on Fr Feb 17 2006 - 17:24:12 CET

This archive was generated by hypermail 2.3.0 : Di Dez 05 2023 - 07:15:07 CET