I can try to strip then all out and see what happens.

That’d be good!  With a Note to explain why…

We still need ctxt_prec, I think, to pass to the ordinary type pretty-printer, when we have HsCoreTy.

Simon

From: Alan & Kim Zimmerman [mailto:alan.zimm@gmail.com]
Sent: 26 May 2017 12:14
To: Simon Peyton Jones <simonpj@microsoft.com>
Cc: ghc-devs@haskell.org
Subject: Re: HsParTy

When I was working through the ppr to use HsParTy everywhere it was not clear whether there were any usages that required maybeParen, so I left it in.

I can try to strip then all out and see what happens.

Alan

On 26 May 2017 at 10:54, Simon Peyton Jones <simonpj@microsoft.com> wrote:

Alan

Do you know why HsTypes.ppr_mono_ty contains any uses of ‘maybeParen’?  Shouldn’t the parens all be put in by HsParTy?

I tripped over this when I mistakenly added what I thought were missing maybeParen calls for HsAppTy and HsAppTys, but that was flagged as an error by a test in tests/printer.

So why are there /any/ maybeParen calls?

Simon