* Re: Text embedded in Bl
2013-09-22 22:33 Text embedded in Bl Anthony J. Bentley
@ 2013-09-22 23:42 ` Ingo Schwarze
0 siblings, 0 replies; 2+ messages in thread
From: Ingo Schwarze @ 2013-09-22 23:42 UTC (permalink / raw)
To: discuss; +Cc: Anthony J. Bentley
Hi Anthony,
Anthony J. Bentley wrote on Sun, Sep 22, 2013 at 04:33:21PM -0600:
> Consider the following, based on an example from gnugk(1):
I saw the commit message on ports-changes@ and thought, "ugh, this
looks fishy, i should probably look into it" but wasn't sure to find
the time. Thank you for looking so carefully into mandoc matters.
This is really helpful.
> .Bl -tag -width Ds
> This is an opening paragraph.
> .It Fl a
> Perform behavior foo.
> ...
> .El
>
> Mandoc refuses to display the page:
> FATAL: child violates parent syntax
Oh dear.
Right, that crash is still in the code.
Now that you talk about it, i see that it's not even listed as a bug.
That one went right to the top of the TODO list, i consider it the
worst bug currently known, as it's a needless crash.
> Groff, on the other hand, displays it (I'm not sure of the specifics,
> but my guess is it treats the first part as a Pp):
>
> This is an opening paragraph.
>
> -a Perform behavior foo.
>
> To me, this is obviously wrong. To realize why, just think of the case
> where you want two paragraphs. (Or the case where the Bl only contains a
> single paragraph with no items--which also appeared in this manpage!)
> Based on this reasoning, I sent a fixed manpage to upstream, who committed
> it.
Yes, sure it's a syntax error, and sure it should be fixed upstream,
thank you for having upstream fix the mdoc(7) syntax error.
> But now I wonder. Should mandoc mimic groff here? I think no,
> but maybe someone else will have a different opinion.
Of course it should.
The aim is that mandoc be bug-compatible with groff.
Well, unless it is terribly difficult to implement, or clobbers the
code with a ton of spaghetti, or groff is so blatantly broken that
i prefer to go and fix groff. But the case at hand, what groff does
is quite reasonable: When the markup makes no sense whatsoever,
just show the text without the markup.
In any case, mandoc mustn't crash, but keep going.
Yours,
Ingo
----- 8< ----- schnipp ----- >8 ----- 8< ----- schnapp ----- >8 -----
Date: Sun, 22 Sep 2013 19:21:23 -0400 (EDT)
From: schwarze@mdocml.bsd.lv
To: source@mdocml.bsd.lv
Subject: mdocml: Throwing FATAL MANDOCERR_SYNTCHILD on non-.It children of .Bl
Log Message:
-----------
Throwing FATAL MANDOCERR_SYNTCHILD on non-.It children of .Bl
feels so 2008-ish, one of the last remnants of the
Kristaps-doesnt-like-syntax-errors-so-lets-just-give-up--Era.
Modified Files:
--------------
mdocml:
TODO
Revision Data
-------------
Index: TODO
===================================================================
RCS file: /usr/vhosts/mdocml.bsd.lv/cvs/mdocml/TODO,v
retrieving revision 1.155
retrieving revision 1.156
diff -LTODO -LTODO -u -p -r1.155 -r1.156
--- TODO
+++ TODO
@@ -7,7 +7,13 @@
* crashes
************************************************************************
-None known right now.
+- .Bl -tag followed by a text node preceding the first .It should not
+ throw a FATAL error, but only a normal ERROR. Putting this into the
+ HEAD of an implicit .It might be cleanest, inserting an implicit .Pp
+ or just dumping the orphan stuff directly into the BODY of the .Bl
+ might be easier to implement, and all options can no doubt be made
+ to yield correct (i.e. groff bug-compatible) rendering.
+ Anthony J. Bentley on discuss@ Sun, 22 Sep 2013 16:33:21 -0600
************************************************************************
* missing features
--
To unsubscribe send an email to source+unsubscribe@mdocml.bsd.lv
--
To unsubscribe send an email to discuss+unsubscribe@mdocml.bsd.lv
^ permalink raw reply [flat|nested] 2+ messages in thread