[9front] [PATCH] troff: we are not htmlroff

[9front] [PATCH] troff: we are not htmlroff

[Humm]

If we don’t explicitly check for ‘h’ in troff, we can’t reliably check
for non-htmlroff well.

Consider the following:

	.if h \{\
	.	de M
	.		tm m
	..\}

Without this change, this will print m and not define macro M.
---
diff 24b9a78ceb63234ef0b0e7d1da0d46119b7a297d c2f89dc6c142284fc3b4c38a4e9fbf5c0c5b2a1d
--- a/sys/src/cmd/troff/n5.c	Wed Nov 24 15:41:36 2021
+++ b/sys/src/cmd/troff/n5.c	Wed Nov 24 20:56:44 2021
@@ -683,6 +683,7 @@
  		if (TROFF)
  			true++;
  		break;
+	case 'h':
  	case ' ':
  		break;
  	default:

Re: [9front] [PATCH] troff: we are not htmlroff

[Humm]

ping

Anyone willing to look at this, to commit this, to send nice words, to 
send any reply at all?

-- 
Humm

Re: [9front] [PATCH] troff: we are not htmlroff

[Sigrid Solveig Haflínudóttir]

I tried applying this but all I get is

applying troff: we are not htmlroff
/rc/lib/rcmain: patch failed
/bin/git/import: rc 190991: patch failed

Can you perhaps git/export and send it as an attachment? Thank you.

Re: [9front] [PATCH] troff: we are not htmlroff

[Humm]

[-- Attachment #1: Type: text/plain, Size: 409 bytes --]

Quoth Sigrid Solveig Haflínudóttir:
>I tried applying this but all I get is
>
>applying troff: we are not htmlroff
>/rc/lib/rcmain: patch failed
>/bin/git/import: rc 190991: patch failed
>
>Can you perhaps git/export and send it as an attachment? Thank you.

Sure thing.  I suppose NeoVim, which I use to compose emails, thought 
it had to double the spaces at the beginning of lines in the diff.

-- 
Humm

[-- Attachment #2: troff.diff --]
[-- Type: text/plain, Size: 645 bytes --]

From: Humm <hummsmith42@gmail.com>
Date: Wed, 24 Nov 2021 19:56:44 +0000
Subject: [PATCH] troff: we are not htmlroff


If we don’t explicitly check for ‘h’ in troff, we can’t reliably check
for non-htmlroff well.

Consider the following:

	.if h \{\
	.	de M
	.		tm m
	..\}

Without this change, this will print m and not define macro M.
---
diff 24b9a78ceb63234ef0b0e7d1da0d46119b7a297d c2f89dc6c142284fc3b4c38a4e9fbf5c0c5b2a1d
--- a/sys/src/cmd/troff/n5.c	Wed Nov 24 15:41:36 2021
+++ b/sys/src/cmd/troff/n5.c	Wed Nov 24 20:56:44 2021
@@ -683,6 +683,7 @@
 		if (TROFF)
 			true++;
 		break;
+	case 'h':
 	case ' ':
 		break;
 	default:

Re: [9front] [PATCH] troff: we are not htmlroff

[Sigrid Solveig Haflínudóttir]

Thanks, that worked.

On Sun, Dec 5, 2021 at 2:30 AM Humm <hummsmith42@gmail.com> wrote:
>
> Quoth Sigrid Solveig Haflínudóttir:
> >I tried applying this but all I get is
> >
> >applying troff: we are not htmlroff
> >/rc/lib/rcmain: patch failed
> >/bin/git/import: rc 190991: patch failed
> >
> >Can you perhaps git/export and send it as an attachment? Thank you.
>
> Sure thing.  I suppose NeoVim, which I use to compose emails, thought
> it had to double the spaces at the beginning of lines in the diff.
>
> --
> Humm

Re: [9front] [PATCH] troff: we are not htmlroff

[Noam Preil]

Hey,

htmlroff(6) says to use `.if !h .if t` to check for normal troff. What's
the issue with that approach?

- Noam Preil

Re: [9front] [PATCH] troff: we are not htmlroff

[Noam Preil]

Ah hm, I think I misunderstood.

The issue is that we don't correctly detect, in normal roff, that we're
NOT htmlroff, not the inverse? Is this related to the differing behavior
with conditionals between troff and htmlroff? 

- Noam Preil

Re: [9front] [PATCH] troff: we are not htmlroff

[Humm]

Quoth Noam Preil:
>The issue is that we don't correctly detect, in normal roff, that 
>we're NOT htmlroff, not the inverse?

Correct.

In normal troff, we don’t know that `h` can be used as condition in an 
`.if` request.  What we perhaps should do is to handle any character 
we don’t know as false.  What we do instead is weird and wonky 
unexpected behavior.  Some stuff gets executed when it shouldn’t.  
Some stuff doesn’t get executed when it should.  The current troff 
behavior on `.if h` and `.if !h` is complete garbage.

What this patch does is not to improve troff’s general behavior on 
unknown conditions, but to let it know `h` is false.

>Is this related to the differing behavior with conditionals between 
>troff and htmlroff?

	; troff -a # same with htmlroff
	.de xx
	.tm a
	..
	.if 0 \{\
	.de xx
	.tm b
	..
	.\}
	.xx
	a       ← output

I guess htmlroff(6) is wrong.

-- 
Humm

Re: [9front] [PATCH] troff: we are not htmlroff

[Noam Preil]

I'd rather we fixed conditional evaluation in general, but it's probably
worth applying this in the meantime anyways.

- Noam Preil

Re: [9front] [PATCH] troff: we are not htmlroff

[Humm]

Quoth Noam Preil:
>Hey,
>
>htmlroff(6) says to use `.if !h .if t` to check for normal troff. What's
>the issue with that approach?

It does not work.  If `.if h` would work correctly, that would check 
for normal troff as opposed to both htmlroff and nroff, but as it is, 
it does not work.  Try:

	; troff -a
	.if !h .if t \{\
	.de M
	.tm m
	       ← output
	m      ← output

Not what you expect at all: Yes, this is troff, but we should be 
defining a macro.  The `.tm` should not get executed, and there should 
not be a blank line before its output.

-- 
Humm

Re: [9front] [PATCH] troff: we are not htmlroff

[ori]

Quoth Humm <hummsmith42@gmail.com>:
> 
> If we don’t explicitly check for ‘h’ in troff, we can’t reliably check
> for non-htmlroff well.
> 
> Consider the following:
> 
> 	.if h \{\
> 	.	de M
> 	.		tm m
> 	..\}
> 
> Without this change, this will print m and not define macro M.
> ---
> diff 24b9a78ceb63234ef0b0e7d1da0d46119b7a297d c2f89dc6c142284fc3b4c38a4e9fbf5c0c5b2a1d
> --- a/sys/src/cmd/troff/n5.c	Wed Nov 24 15:41:36 2021
> +++ b/sys/src/cmd/troff/n5.c	Wed Nov 24 20:56:44 2021
> @@ -683,6 +683,7 @@
>   		if (TROFF)
>   			true++;
>   		break;
> +	case 'h':
>   	case ' ':
>   		break;
>   	default:
> 

Generally looks ok to me, but I'm not a troff expert.
Was hoping someone a bit more knowlegable would look,
but I'm willing to commit if nobody else pipes up.

thanks for pinging.