perm filename ANDREI.LE1[LET,JMC] blob sn#042005 filedate 1973-10-04 generic text, type C, neo UTF8
COMMENT āŠ—   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\{\M0CYR30\M1NGR30\M2NGR40\M3BDI25\M4XMAS25\M5FIX13\M6CLAR30}\.
C00008 ENDMK
CāŠ—;
\{\M0CYR30;\M1NGR30;\M2NGR40;\M3BDI25;\M4XMAS25;\M5FIX13;\M6CLAR30;}\.
                 {\F2ARTIFICIAL INTELLIGENCE LABORATORY
                     COMPUTER SCIENCE DEPARTMENT
                     STANFORD, CALIFORNIA 94305}



                                {\F110} mart 1973g.



Professoru A. P. Ershovu
Akademgorodok

                                Dorogoyj Andreyj!

{\J}	Kak  Vam  nravitsya  nasha  novaya  shiroko-pechatj?      Ona
nazihvaetsya {\F3XGP (Xerox Graphics Printer)} i sdelana firmoyj {\F3Xerox}
na osnovanii mashinih {\F3LDX}.   Centr  isledovaniya  {\F3Xerox}  v  {\F3Palo
Alto} sdala nam mashinu bezplatno, chtobih issledovatj ee primeneniya.
{\F3XGP}  poluchaet  iz  CVM   do   1700   razryadov   kazhdihyj   pyatj
millisekundov. Ehto {\F3scan line} kak po televidenii.   Bukvih sdelanih
iz tochek programmoyj. Stranica sostoit iz 1700{x}2200 {=}  3,740,000
razryadov.  Iz-za  ehtogo, CVM dolzhna rabotatj ochenj bihstro.  Ili
ona dolzhna imetj v operativnoyj pamyatj okolo 100,000 slov,  ili  ona
dolzhna   poluchatj   iz   diska   dovoljno   boljshoyj  bufer  ochen
regulyarjno,  potomu  chto,  kogda   bumaga   nachinaet   dvigatj   v
shiroko-pechati, ona ne mozhet ostanovitjsya do 22 dyuzhinihx.  Principe
mashinih - takoyj zhe kak u obihknovennoyj mashinih {\F3Xerox}.

	Kak  Vih   vidite,   mashina   ochenj   gibka.      Vozmozhno
polzovatjsya  lyubihm  alfavitom,  v  lyubom razmere, i krome togo,
vozmozhno pechatatj illustracii.  
Naverno Vih tozhe zametili, xoroshaya shiroko-pechatj ne pomogaet
moemu ploxomu russkomu yazihku.
Seyjchas vernyusj  na  angliyjskiyj
yazihk.{

	The  XGP  only  started working a month ago, and we are still
collecting different sizes and shapes  of  characters  for  it.   The
Russian   font  used  in  this  letter  was  photographed  using  the
television camera from the alphabet at the beginning of a textbook of
Russian.  We  would like to get a Russian font book, i.e. a book with
many alphabets in different sizes and styles.  We hope to be able  to
get  printing  almost  of book quality from the printer.  It has also
been used to print logic diagrams from the foonly project.

	It looks like the production documentation for foonly will be
completed  in  another month, but the government has lost patience so
that production is  uncertain.   At  least  one  commercial  firm  is
interested in producing the machine.

	While  I  was  at M.I.T., I did not make much progress in the
writing I was going to do, so I have decided not to make any  foreign
trips this year.

	We  received a film from Okhotsimsky and have sent him one of
ours in return and also a recent thesis he requested.

	We are looking  forward  to  seeing  you  at  the  Artificial
Intelligence  Conference.   If  you  can  come early or stay late, we
would be glad to have you as our guest. \.


					Best Regards,



					\F4John


\F5\J	P.S. I enclose the source document of this letter.  The
Russian part is written in a version of latinized Russian that has
the property that any combination of Russian letters is uniquely
representable, not just the ones that adhere to Russian orthography.
The source text then goes through a transliteration program that uses
a transliteration table.  Then it goes through the justification
program that knows the widths of the characters.  Finally, it goes
to the XGP print program in the system.  The transliteration program
will hopefully work for the Japanese kana letters, Greek, Hebrew, etc.

	I am presently dividing my time among my book on technology,
mathematical theory of computation, and representation but have no
startling results to report in any of these areas. \.