[comp.text.tex] TeX and 1270 dpi Linotronic output

tj@gpu.utcs.utoronto.ca (Terry Jones) (05/26/90)

We have a Linotronic typesetter at University of Toronto and I would
be interested in hearing about user experiences with TeX and high 
resolution output devices.

In particular jobs fail regularly with a VM Error (out of memory) and
I am interested in hearing of others with similar experience and of ANY
possible solutions to this. 

1270 dpi fonts, aside from eating disk space like chocolate chip cookies
are cumbersome to move around networks, and consume memory on the linotronic.
This causes most of the problems as far as I can determine. Some pages
when processed with a dvi2ps that uses the builtin fonts on the lino for
all but math STILL failed when there were a lot of different fonts on
a page (8 different families or sizes). This appeared to be very inefficient
re-encoding of the builtin fonts. 

Lets hear from anyone please...

tj

jansteen@cwi.nl (Jan van der Steen) (05/28/90)

In comp.fonts you write:

>We have a Linotronic typesetter at University of Toronto and I would
>be interested in hearing about user experiences with TeX and high 
>resolution output devices.

Well, I don't have experience with TeX in particular but I do have
experience with the Linotronic in combination with downloaded fonts.
In particular, we have a collection bitmap fonts in PostScript format
at 720 dpi used to print house style publications. When processing
those jobs on the Linotronic we also did run out of memory.

>In particular jobs fail regularly with a VM Error (out of memory) and
>I am interested in hearing of others with similar experience and of ANY
>possible solutions to this. 

The solution I came up with was to write some kind of "garbage collector".
The garbage collector will do a "vmstatus" each time a new font
is created. If the memory drops down a certain value (I took as
threshold 250000 bytes) all important environment values are written to
disk and a "restore" of a previous "save" is performed.
After the "restore" the environment (which is erased by a "restore"
action) is fetched from disk and we can continue with a clean memory.
People wo are interested in the code can mail me.

>1270 dpi fonts, aside from eating disk space like chocolate chip cookies
>are cumbersome to move around networks, and consume memory on the linotronic.
>This causes most of the problems as far as I can determine. Some pages
>when processed with a dvi2ps that uses the builtin fonts on the lino for
>all but math STILL failed when there were a lot of different fonts on
>a page (8 different families or sizes). This appeared to be very inefficient
>re-encoding of the builtin fonts. 

>Lets hear from anyone please...

I talked to the Linotronic salesman about this problem and he inquired
the research lab about any memory extension possibilities.
There weren't any! However, they recognized the problem and may do
some modifications on the ROM's to support more memory in future releases.

The limited memory of the Linotronic made us decide to buy a Compugraphic
typesetter driven by an Agfa engine with Adobe ROM's. This implementation
supports 4 Mb of virtual memory; enough for our purposes.


	Jan van der Steen
--
    -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-
     Jan van der Steen                 jansteen@cwi.nl
     Centre for Mathematics and Computer Science (CWI)
     Kruislaan 413, 1098 SJ Amsterdam, The Netherlands

mario@madarch.man.ac.uk (Mario Wolczko) (06/06/90)

In article <1990May26.040231.22931@gpu.utcs.utoronto.ca>,
tj@gpu.utcs.utoronto.ca (Terry Jones) writes:
> We have a Linotronic typesetter at University of Toronto and I would
> be interested in hearing about user experiences with TeX and high 
> resolution output devices.
> 
> In particular jobs fail regularly with a VM Error (out of memory) and
> I am interested in hearing of others with similar experience and of ANY
> possible solutions to this. 

We use a Linotronic with a modified version of LaTeX, which I call
PS-LaTeX (it uses PostScript fonts where possible).  It's 99.9% LaTeX
compatible.  The use of PostScript fonts circumvents most of the VM
problems (we only have trouble with LaTeX picture mode, which means we
have to do those pages separately).  Otherwise, works fine.  For the
proof, look at these two books, published in the last 6 months:

  C.B.Jones, Systematic Software Development using VDM, 2nd ed,
  Prentice-Hall International, and
  C.B.Jones and R.C.F.Shaw, eds, Case Studies in Systematic Software
  Development, Prentice-Hall International

pslatex can be got from the Clarkson server (sun.soe.clarkson.edu).

Mario Wolczko

   ______      Dept. of Computer Science   Internet:      mario@cs.man.ac.uk
 /~      ~\    The University              USENET:    mcsun!ukc!man.cs!mario
(    __    )   Manchester M13 9PL          JANET:         mario@uk.ac.man.cs
 `-':  :`-'    U.K.                        Tel: +44-61-275 6146  (FAX: 6280)
____;  ;_____________the mushroom project___________________________________

tj@gpu.utcs.utoronto.ca (Terry Jones) (06/07/90)

In an original posting I asked about pro and con comments on
Linotronic output. I should have been more specific. 

We have a Linotronic L100 and I am very interested in hearing about
ALL experiences with people trying to produce tex output on an L100.

The responses I got so far ar all "yes but we use a Linotronic with
GOBS of RAM" (Not an option on the L100) or "use modified drivers
that use the builtin fonts" which works better on the L100 but still we
often have to split the job into 1 page chunks. SOme modified dvips 
programs I gather will help for this.

We are looking into getting the CM fonts in PostScript form to put on
the Lino disk and I hope this will solve MOST of the problems but
I am still wanting to hear about the good AND the bad and ugly from
people about their experiences, especially L100 users!

tj

dhosek@hmcvax.claremont.edu (Hosek, Donald A.) (06/17/90)

In article <1990Jun16.182800.26891@zorch.SF-Bay.ORG>, xanthian@zorch.SF-Bay.ORG (Kent Paul Dolan) writes...
>In article <1990May26.040231.22931@gpu.utcs.utoronto.ca>,
>tj@gpu.utcs.utoronto.ca (Terry Jones) writes:
>> We have a Linotronic typesetter at University of Toronto and I would
>> be interested in hearing about user experiences with TeX and high 
>> resolution output devices.

>> In particular jobs fail regularly with a VM Error (out of memory) and
>> I am interested in hearing of others with similar experience and of ANY
>> possible solutions to this. 

>There seem to be a growing number of similar comments about TeX being
>memory bound (the same problem affects METAFONT); as the resolution of
>display devices continues to increase, existing implementations of TeX
>are failing with lack of memory or lack of array size error messages.

>Does the new release of TeX and METAFONT address these problems?  If not,
>is anyone going through the code looking at ways to overcome these limits?

Well, TeX doesn't know output resolution from Adam, so I really
don't see how there can be a problem with TeX's memory as a
result of output device resolution. Document complexity, perhaps,
although 32bit TeX (the most common version and what you get on
most dinky computers (e.g., PCs, Macs, etc.)) can handle more
complicated documents than almost anything else and the only
cases where one runs into problems is with documents that use
graphics packages like PiCTeX which actually do the graphics
typesetting in TeX itself.

MF also does not maintain a bitmap in memory (at least not all of
the time) so it does a pretty good job of keeping itself going at
decent resolutions. The only times I've seen MF run out of memory
is in the case that MF is used to manipulate bitmaps.

Both of these problems can be solved by switching to a version of
TeX (or MF) which uses 64bit memory words thus giving 32bit
addressibility in the main memory array.

The more likely location for memory problems at high resolutions
is in the printer driver or the printer itself. The Linotronic
typesetters have ridiculously small amounts of VM causing
difficulties with bitmapped fonts. However, intelligent managing
of the font memory (cf. Tom Rokicki's dvips) deals with the
problem rather well.

Also, Type 1 PS versions of CM will probably be a reality in the
next year or so and a version of MF outputting outline fonts will
be available within that same time frame. (Please don't send me
with mail on this latter item--every mail message that I get
delays my posting of the details on the state of the matter by a
little bit more).

-dh

---
Don Hosek                         "Other people get into occupations by 
dhosek@ymir.claremont.edu          accident or design; but writers are born. 
dhosek@ymir.bitnet                 We have to write. I have to write. I 
uunet!jarthur!ymir                 could work at selling motels, or slopping 
                                   hogs, for fifty years, but if someone asked 
                                   me my occupation, I'd say writer, even if 
                                   I'd never sold a word. Writers _write_. 
                                   Other people _talk_."      -W.P. Kinsella