[Haskell-cafe] What features should an (fictitious) IDE for Haskell have?

[Joachim Durchholz]

That's one of the ideas that come up on a regular basis, but it doesn't 
buy you anything in practice.

You want to be able to represent code in mails, websites, and books.
And you want that representation to be unambiguous, so you need to 
define the whole parsing process.

In the past, having a binary representation would help with refactoring 
and such, but modern CPUs are fast enough to create the AST on the fly. 
Even in the presence of syntax errors.

Also, pure AST editors tend to be clunky. They disallow invalid syntax, 
so the programmer is forced to do everything right. There are many 
situations where you start writing down something vague and refine it, 
dealing with those aspects first that you need.
In a sense, the problem is that an AST doesn't give you a scratchpad for 
ideas and vagueness, everything has to be perfect from the get-go.

I first heard about such an idea in the 90s (when on-the-fly parsing 
wasn't really doable). It was one of the things MS wanted to do, and 
they even created a prototype. It failed.

Am 01.12.20 um 02:01 schrieb MarLinn:
> Most importantly: A good IDE is not a text editor, but an AST editor. If 
> the AST happens to be presented as text, that's a choice of 
> visualisation, nothing more. Better to start with a graph-like 
> visualisation to free the mind, then think through the possible 
> interactions. Maybe add the typical text-like visualisation later. But 
> don't start there or you'll just re-invent notepad for the nth time.
> 
> Maybe don't even /store/ the code as ascii text.
> 
> 
> 
> _______________________________________________
> Haskell-Cafe mailing list
> To (un)subscribe, modify options or view archives go to:
> http://mail.haskell.org/cgi-bin/mailman/listinfo/haskell-cafe
> Only members subscribed via the mailman list are allowed to post.
>