X-VM-v5-Data: ([nil nil nil nil nil nil nil t nil]
	[nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil nil])
Date:         Fri, 23 Feb 90 17:10:00 PST
Reply-To: LaTeX-L Mailing list <LATEX-L@DHDURZ1.bitnet>
Sender: LaTeX-L Mailing list <LATEX-L@DHDURZ1.bitnet>
From: DHOSEK@HMCVAX.bitnet
Subject:      Verification
To: Rainer Schoepf <BK4@DHDURZ1.uucp>
Status: R
X-Status: 
X-Keywords:                  
X-UID: 29

>My most serious misgiving about this whole discussion is that I think
>that it displays a lack of perspective.  There are a small number of
>people who are going to be doing the coding, and a lot of things that
>are more important than this very minor enhancement.  Let me suggest
>one such item that I would much rather see people thinking about than
>this sort of trivia.  When writing LaTeX, there were two important
>things that I did not attempt to do properly for lack of time:
>documentation and testing.  Frank and Rainer have invested considerable
>energy into fixing the first problem; I have seen no plans for fixing
>the second.  Version 2.09 has become much more reliable since it was
>introduced in 1985.  People are not going to be very happy if Version
>2.10 takes them back five years in reliability.  Anybody want to
>tackle this problem?

The "trip test" approach to LaTeX is possibly the only workable
approach to verification. Essentially, we need to have documents
that try every supported feature in every meta-style and with
every possible combination of supported options. Just with what
we have no, this comes to a lot of combinations. And it's not
enough to just, say, have a footnote and a marginpar somewhere in
the document (this is a genuine bug that I believe is still in
existance), but to have a footnote immediately after and
immediately before the marginpar (regarding the aforementioned
bug, one of the two combinations gives odd output). The problem
is, we can't merely take the output and "diff" it against a
standard output like we can for TeX; someone needs to look at the
output and say, "yes, that worked." I don't know whether a
complete "trip" is possible. A partial one is possible, and not
that difficult (my in-progress book on LaTeX (developed from my
class notes) has served that purpose to a degree... while
preparing it, I found around half a dozen bugs which Leslie may
remember hearing about).

-dh