Date: Fri, 14 Jul 2000 11:08:31 -0500 (EST)
From: Clarence Wilkerson <wilker@math.purdue.edu>
Subject: Remarks for Hopf users

I. Technical problems

   I have not yet been able to automate the processing of submissions
   done via e-mail to same degree achieved for ftp submissions.

As a sample,

I have had to spend quite a bit of time recently dealing with attachments
to mail generated on Mac systems. Some points

1) The attachments often use a binhex type coding instead of the more
   common uuencode or "base 64" format. There are unix based decoders
   (hexbin ) for this, and I have one.

But ..

2) The file produced by the decoder often includes
   a) a header ( often 128 bytes ) , probably containing info
      about the resource fork of the file.
   b) then the data fork  ( this is the only thing Hopf needs )
   c) extra stuff ( padding?) at the end.

It would save me trouble if the one the data forks are included
in the attachments. I'm looking for a unix solution ( I think
( hexbin -d   attachment.bin, where attachment.bin is produced
  by metamail ) should work ). If any of you mac/unix uses have
hints on this problem, please let me know.

I've notice in the past similar problems in files ftp'ed from a Mac
to a unix system ( both forks included ).


II. Other problems users create!

I continue to get a wide variety of inputs even via FTP.
The most common variance is in the abstract
file. The format I need plain ASCII text,
with name  "mygreatestthm.abstract" corresponding to the DVI
file "mygreatestthm.dvi"  . Not submitting an abstract at all
makes me attempt to produce an approximation from the dvi file
or ps-file. Submitting a proper abstract with a name that doesn't
conform to "mygreatestthm.abstract"  chews up my
time, and in the future will not be  ameanable to more automatic
processing.

I do ( reluctantly) accept submissions in .ps format. However
that I feel that this and other non-DVI formats
hinder easy online accessiblity to their mathematics. I prefer
to offer the  variety of formats that the DVI files afford Hopf.

III. Upcoming changes in Hopf:

I apologize again for the delays this year
created in part by technical problems and my 12 weeks away
from Lafayette.

The changeover to a departmental server for Hopf is still planned.
It will not involve a change in the name hopf.math.purdue.edu,
but the IP number which your DNS server produces will change.
If you have 128.210.3.18 hardcoded into your scripts, you will
have to change it when the switch happens.
Stay tuned.


Clarence