Xref: utzoo comp.text:3307 comp.misc:5186
Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!pacbell!att!homxb!houxa!edwards
From: edwards@houxa.ATT.COM (D.LEWAN)
Newsgroups: comp.text,comp.misc
Subject: Re: HELP! with boxed tables using MM, tbl
Summary: bug in documentation not tbl(1)?
Keywords: MM tbl
Message-ID: <4801@houxa.ATT.COM>
Date: 17 Feb 89 15:06:30 GMT
References: <946@mtuxj.att.com> <1085@hsi.UUCP> <8909@alice.UUCP>
Distribution: usa
Organization: AT&T BL Holmdel NJ USA
Lines: 51

In article <8909@alice.UUCP>, debra@alice.UUCP (Paul De Bra) writes:
> In article <1085@hsi.UUCP> hecht@hsi.UUCP (Alice Hecht) writes:
> ?In article <946@mtuxj.att.com>, btwomey@mtuxj.att.com (XF3AA-W.TWOMEY) writes:
> ?> 
> ?> I'm trying to create multi-page boxed tables using tbl and mm and troff.
> ?> I'm using .TS H and .TH to get the header repeated on pages.
> ?> 
> ?> MM has problems though. I have had: the sides of the box go missing;
> ?> pages come out blank except for the table header; 
etc.................
> ?I've had similar problems with tables before and the only thing that 
> ?ended working was ending the table at the bottom of each page and then
> ?restarting it again at the top. 
etc.................
> 
> From working with people that are serious troff/tbl hackers I can predict
> that the bug most likely is in tbl, not in MM or troff. Tbl never checks
> the distance to the next trap, 
etc.................

I will not advocate tbl(1) as perfect -- far from it -- but the
problem described is anticipated in both the tbl(1) documentation
and implementations.  To quote:

	The macro T# is defined to produce the bottom lines and side
	lines of a boxed table, and then invoked at its end.  By the
	use of this macro in the page footer a multi-page can be
	boxed.

It works -- generally, certainly not always; it seems that when
things happen too close to traps troff gets confused about which
side they're on and messes up accordingly.

The bug is that the detail is so hidden in the documentation, and
that is a side-effect of the fact that the documentation is so
unstructured and scattered.  It looks just like a research project
that's been artificially elevated to the status of product without
regard for the non-academic features that make *products* reasonable
for use by people.  Academic projects uncover universal truths;
their results live independently.  

Any other favorite bugs in UNIX\(rg text processing?  My favorites
come from the fact that it is all interpreted and chained and there
is no way to truly integrate.  Others come from the 2 character
restrictions on names.  Oh, yeah, most tbl(1) implementations allow
only 100 text blocks.  C'm'on guys, in computer science is 100
really such a large number?????

Yours in box rules,

Doug