[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Lout 3.17 runs not converging

From: Valeriy E. Ushakov
Subject: Re: Lout 3.17 runs not converging
Date: Thu, 2 Dec 1999 21:34:29 +0300

On Thu, Dec 02, 1999 at 04:40:15PM +0000, Albert Kinderman wrote:

> I am running Debian unstable (potato) and compiled lout 3.17 from
> source.  "Slides" converges in two runs.  The user guide converges
> in 4 runs.

I tested on the following x86 boxes:

    Win32               VC6/sp3         -Ox (maximum optimization)

    FreeBSD-3.3         gcc-     -O2

    RedHat-6.1          egcs-2.91.66    -O3
    2.2.12-20smp        libc-2.1.2

and on SS20 with

    Solaris-2.6         egcs-2.91.60    -O

On all these systems I get identical results.  `slides' happily
converges in two runs.  `user' converges in 6 runs.

For `user' on x86 boxes after the 5th run I get:

    gra_summ:8:1: 23.6c object too high for 23.6c space; @Scale inserted
    all:14:1: unresolved cross reference @Runner&&following
    all:14:1: unresolved cross reference @Runner&&following

where the first one is really: 23.57848c object too high for 23.57672c.

Diffing xref db from 5th run produced on FreeBSD and RedHat shows no diffs.

After the 6th run document converges.  Diffing all the three xref dbs
and the xref db from Sparc shows no diffs (well, for win32 it does
show diffs, but those are due to CRLF line ends).  Ediffing postscript
output shows only occasional minor deviations in numbers.  Ediffing
with Sparc output shows more deviations but nothing significant.

PS: Are you sure `user' converged in 4 runs?  Does the 5th run produce
    identical results?  4th run on Sparc produces no warnings, but it
    has not converged yet - save lout.li *.ld and *.ps run again and
    diff to see.

PPS: Emacs' ediff-files (under X or Win32) is really nice way to check
    where PostScript ouputs differ as ediff will refine differences to
    word granularity - which is much easier than hunting for 7192 vs
    7193 in diff output.

PPPS: It might be a good idea to strut running headers vertically , so
    that the height of headers does not depend on the running title
    (e.g. ascenders, descenders) - though I haven't tried it.

SY, Uwe
address@hidden                         |       Zu Grunde kommen
http://www.ptc.spbu.ru/~uwe/            |       Ist zu Grunde gehen

reply via email to

[Prev in Thread] Current Thread [Next in Thread]