[Haskell-cafe] Folded long string literals and CPP?

[Jeff Clites]

I like it. I would be interesting (though perhaps overkill) to have a special operator which is like ++ but which can only be applied to list literals (not only strings), and which is guaranteed to be syntactic sugar for the concatenated list. One step fancier would be to allow it not just for literals but for any known-at-compile-time values. (So it’s syntactically an operator but in actuality more of a compiler directive, to evaluate the concatenation at compile time, or fail if that’s not possible.)

Just thinking out loud.

Jeff

> On Nov 21, 2023, at 5:14 PM, Richard O'Keefe <raoknz at gmail.com> wrote:
> 
> A tiny change to Haskell, borrowed from Erlang, would solve the problem,
> I think, with an increase in readability.
> 
> In Erlang, "c1...cn" ++ Pattern is a pattern, equivalent (in Haskell terms) to
> ('c1' :: 'c2' :: ... 'cn' :: Pattern).  This means that a long string
> can be written
> as ("..." ++ "..." ++ ... ++ "..."), spread across as many lines as you need,
> with each substring being complete on a line, in any context you want.
> In an expression context, you can already do this, and that is easier to
> read than strings broken across multiple lines. It makes the treatment of
> leading white space completely obvious (if it's inside string quotes, it's
> part of the text, if not, it's not).  The only problem is that you cannot use
> that construction where a pattern is needed (although if you are trying to
> use a long string as a pattern I have to wonder why).  So the tiny
> extension is to allow an explicit concatenation of string literals as a string
> literal in a pattern.
> 
> Right now,
> 
> hello :: String
> hello = "Hello\ \
>          \ World!"
> 
> can and should be written
> 
> hello :: String
> hello = "Hello " ++
>            "World!"
> 
> 
>> On Tue, 21 Nov 2023 at 23:53, Viktor Dukhovni <ietf-dane at dukhovni.org> wrote:
>> 
>>> On Tue, Nov 21, 2023 at 09:27:12AM +0000, Dan Dart wrote:
>>> 
>>> That does sound like a bug. but it doesn't seem patchable using {-#
>>> LANGUAGE CPP #-} because it *is* a bug in -XCPP... :p
>>> 
>>> Have you reported it anywhere?
>> 
>> Not yet reported outside this thread.  Sure, morally speaking, this is a
>> bug in "-XCPP", but in practice "-XCPP" unavoidably uses the actual C
>> preprocessor, and is subject to its syntax idiosyncrasies.
>> 
>> There's not a lot GHC can do once CPP joins the split lines and drops
>> the trailing "\", and the documented trailing whitespace work-around no
>> longer (if it once did) helps.
>> 
>> So the only options appear to be to live with editing the code to to use
>> a trailing "\ \" instead of just "\" when enablng -XCPP (and reverting
>> when dropping "-XCPP", or else introduce a new syntax for multi-part
>> string literals.
>> 
>> Unless someone can suggest a better idea...
>> 
>> --
>>    Viktor.
>> _______________________________________________
>> 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.
> _______________________________________________
> 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.