Re: two things I wanted from TH today

"Simon Peyton-Jones" <[email protected]> writes (quoting me):

> | > But it doesn't look like Pattern Guards are available in THSyntax
> yet.
>
> The question is whether we want THSyntax to absorb all GHC extensions.
> I guess the answer does seem to be 'yes, on demand' 
(snip)

Let's say we have a library that exports TH functions, and a user of
that library.

I imagine that the user of TH-generated code can use -fth (but not
-fglasgow-exts) and be safe if the code compiles without complaining:
they're not using any extensions.

But nothing prevents the author of the library from "accidentally"
employing glasgow-exts in TH-generated code... It would be nice if the
compiler could somehow prevent me from doing so.  Some ideas offhand:

- Have another module, THSyntaxGlasgowExts.

- Have two packages provide THSyntax, one has extensions and one
  doesn't.  That way, the user can employ a command-line flag to allow
  or disallow extensions, just as they do for hand-generated Haskell
  code.

In any case, I'd tend to agree that it's fine to add extensions to
THSyntax for now, since TH users probably use extensions, and once TH
settles down more (and maybe gets adopted by other implementations
some day), you can always implement a non-extension API...

> because the kind of people who use TH also seem to be the kind of
> people who use GHC extensions!

I can imagine that the reasons that some people avoid extensions are
that 1) they want to remain compatible across compilers, 2) they are
concerned for the longevity of code that employs unstable language
features, and 3) they don't understand or don't want to understand the
extensions.  Of course, the same reasoning applies to using TH since
it is itself an extension :)


peace,

isaac