[plt-scheme] error message for case-sensitivity

From: Anton van Straaten (anton at appsolutions.com)
Date: Tue May 25 20:31:02 EDT 2004

ifconfig nslookup wrote:
> Anton said:
> > All this highlights the point here: it's messy because even
> > with something as simple as case-insensitivity, we're talking
> > about encoding some arbitrary human knowledge about natural
> > language, into the core of a programming language.
>
> Who said it has to be in the core of a programming language?
> I don't consider the default error handlers the core of a programming
> language.

I was speaking to the broader discussion, of case-sensitivity in general.

> So what is it that's wrong with encoding it into the error
> handler? It may be arbitrary, but we want it because it's
> a common mistake.

Is it, though?  How often does this really happen to you?

The example that triggered the current thread is one where the name
sxml->html had been changed to sxml:sxml->html, and e.g. SRV:send-reply was
changed to sxml:display-fragments.  Case-sensitivity is a red herring in
this case.  An error handler which does a substring search could have
figured out the first one, but nothing short of module-specific *historical*
knowledge would help you with the second.

Besides, I'd rather see a problem like this solved before an error occurs.
Isn't the requirement really for a general-purpose identifier lookup helper,
which you might want to invoke when first typing in the code, or after
running Check Syntax and seeing an identifier highlighted in red, or, worst
case, after an error has occurred?

> P.S.
> I hate Microsoft Outlook.
> Does anyone know how to make it stop changing my nick's i to an I?

A great example of what can happen when machines try to apply heuristics
from natural language!  ;)

Anton



Posted on the users mailing list.