Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!sdd.hp.com!spool.mu.edu!munnari.oz.au!uhccux!uhunix2.uhcc.Hawaii.Edu!julian
From: julian@uhunix2.uhcc.Hawaii.Edu (Julian Cowley)
Newsgroups: comp.text.tex
Subject: Re: TeX bug or feature?
Message-ID: <13133@uhccux.uhcc.Hawaii.Edu>
Date: 21 May 91 09:57:18 GMT
Article-I.D.: uhccux.13133
References: <13021@uhccux.uhcc.Hawaii.Edu> <1991May15.223441.23055@csrd.uiuc.edu> <13067@uhccux.uhcc.Hawaii.Edu> <1991May16.155355.6792@csrd.uiuc.edu>
Sender: news@uhccux.uhcc.Hawaii.Edu
Organization: University of Hawaii at Manoa
Lines: 41

Victor Eijkhout <eijkhout@s41.csrd.uiuc.edu>  writes:
>julian@uhunix2.uhcc.Hawaii.Edu (Julian Cowley) writes:
>>Victor Eijkhout <eijkhout@s41.csrd.uiuc.edu> writes:
>>>julian@uhunix2.uhcc.Hawaii.Edu (Julian Cowley) writes:
>>>>Can anyone explain to me why TeX sets \pagegoal to the g present
>   [...]

>>>The \pagegoal is set to the \vsize when the first nondiscardable item is
>>>added to the current page.  After that it doesn't change, except when an
>  [...]

>>That does make sense, but really what I was after is the height
>>of all insertions present when the \output routine is called.

>You can measure that by just asking for the heights of
>the insertion boxes.

Unfortunately, not while \holdinginserts=1, since TeX doesn't
put anything in the insertion boxes.  What I'm trying to do is
design a two-column mode which allows insertions spanning both
columns at the top of the page.  My thought was that I could
find the height of the double insertions without removing any of
the other insertions from the vertical list (such as footnotes)
by using two passes through the vertical list, one with
\count\doubletopins=1000 and a second with \count\doubletopins=0,
and using the difference of the page goals.  This too has
problems, since on the second pass TeX thinks that it can move
such an insertion onto the page with no cost to the \pagegoal.
If TeX had an easy way to specify which insertions to hold back,
or even a way to get the heights of specific insertions while
\holdinginserts=1, my problems would be over.

I'm now of the conviction that TeX really is inequipped to deal
with double columns, at least with regards to inserts.  It may
be possible to get good results using TeX for general two-column
documents, but I think that insertions spanning more than one
column add a complexity to the page breaking algorithm that TeX
can't handle well.

Julian Cowley / U. of Hawaii at Manoa	| "God is love." -- John, c. 60 A. D.
julian@uhunix.uhcc.hawaii.edu		| "All you need is love." -- John, 1967