00100	HI LEE, THIS IS LARRY TESLER.  YES THAT WAS ME.
00200	WE MET AND AGREED ON A REMARKABLE NUMBER OF POINTS.
00300	THEN JOHN CAME IN AND DID NOT AGREE.  HOWEVER, I AM
00400	SURE WE WILL COME UP WITH A NUMBER OF CONCLUSIONS
00500	EARLY NEXT WEEK.  THEY SEEM SOMEWHAT DIFFERENT FROM
00600	YOURS IF I UNDERSTAND YOURS CORRECTLY.  SO LET ME
00700	TRY TO SELL YOU ON THEM.  THESE ARE THE UNANIMOUS
00800	CONCLUSIONS OF LES EARNEST, BOB SPROULL, LYNN QUAM,
00900	BRIAN HARVEY, AND ME.
01000	
01100	SEVERAL STANDARDS ARE NEEDED.  THEY INCLUDE:
01200		UNFORMATTED DOCUMENT
01300			MIXED TEXT AND GRAPHICS
01400			COMPLETELY EXPLICIT PUB-LIKE CONTROL INFO
01500			MACRO AND PROCEDURE FACILITY
01600				(SAME SYNTAX BUT EXPANSION-TIME
01700					DIFFERENT)
01800			USUALLY REPRESENTED AS HUMAN-READABLE ASCII
01900				FILE TO ALLOW EDITING,
02000				BUT: (1) MAY BE CREATED BY A
02100				PREPROCESSOR FROM SOURCE MANUSCRIPTS
02200				WITH IMPLICIT CONTROL INFO (CF.
02300				"AT" IN PUB); (2) FOR COMPACTNESS,
02400				PARTS MAY BE STORED ON OTHER FILES AND
02500				"REQUIRED"; (3) "REQUIRED" HALF-TONES
02600				AND SUCH MAY BE STORED IN SPECIAL
02700				COMPACT REPRESENTATIONS (USUALLY TO
02800				BE CONVERTED BACK TO ASCII FOR NETWORK
02900				TRANSMISSION OR USER EXAMINATION)
03000			UNPAGINATED!  CROSS-REFERENCES UNRESOLVED!
03100			DEVICE-INDEPENDENT (BUT CONDITIONAL MACROS
03200				AND PROCEDURES CAN HAVE DEVICE
03300				CONDITIONALS)
03400			FONTS REFERRED TO NON-SPECIFICALLY, E.G.,
03500				"FONT 1".  THE EQUATION TO SPECIFIC
03600				FONTS TO BE MADE BY SWITCHES TO THE
03700				FORMATTER (AT THE TOP OF THE FILE THERE
03800				COULD BE DEFAULTS)
03900	
04000		FONT INFORMATION FILE
04100			THIS IS COMPOSED OF ALL FONTS NEEDED
04200			BY A DOCUMENT.  FOR EACH
04300			FONT THERE IS INFORMATION ABOUT THE GEOMETRY
04400			OF EACH CHARACTER (LIKE HEIGHT AND WIDTH)
04500			IN FACT EVERYTHING BUT ITS ACTUAL IMAGE.
04600			SINCE THE F. I. FILE IS SIMPLY A COLLECTION OF
04700			HEADERS FROM FONT FILES, IT FOLLOWS THE
04800			SAME STANDARD AS THOSE HEADERS.
04900	
05000		FONT MASTER FILE
05100			THE COMPLETE DESCRIPTIONS OF THE FONTS
05200			AVAILABLE IN THE SYSTEM.  ACTUALLY THIS IS
05300			SEVERAL FILES.  ONE IS DEVICE-INDEPENDENT;
05400			EACH CHARACTER IS REPRESENTED BY A CLOSED
05500			CURVE OUTLINE WITH SCALING HEURISTICS.
05600			THE OTHERS ARE DEVICE AND SCALE-DEPENDENT --
05700			XGP BIT MATRICES, IMLAC VECTOR COORDINATES,
05800			DATADISC BIT MATRICES, ETC.  SOME
05900			OF THESE CAN BE COMPUTED FROM THE DEVICE-
06000			INDEPENDENT FILE EASILY, BUT OTHERS WILL
06100			REQUIRE CHARACTER EDITING ESPECIALLY ON
06200			LOW BIT-DENSITY RASTER DEVICES.
06300	
06400			EACH CHARACTER IN THE MASTER FILE IS REPRESENTED
06500			BY FOUR COORDINATES (BESIDES THE DEVICE AND SCALING)
06600			(FONT,STYLE,OVERLAY,A) WHERE FONT IS
06700			SAY LYDIAN OR TIMES ROMAN, STYLE IS ITALIC
06800			LIGHTFACE OR BOLDFACE, OVERLAY IS
06900			GREEK, RUSSIAN, SMALL CAPITAL, OR LATIN, AND
07000			"A" IS THE ROOT ASCII CHARACTER.  THIS
07100			SYSTEM WAS USED WITH GREAT SUCCESS AT
07200			COMPOSITION TECHNOLOGY WHERE BRIAN COMES FROM
07300			AS WELL AS MOST OF THESE IDEAS.
07400	
07500			NOTE: MOST ROWS OF THE 4-D ARRAY ARE EMPTY
07600			(E.G., LYDIAN GREEK) AND DEFAULT TO TIMES ROMAN
07700			OR SOMETHING-- HARVEY HAS GOOD DEFAULTING
07800			ALGORITHMS-- SO ITS NOT REALLY A BIG MATRIX.
07900	
08000		FORMATTED DOCUMENT
08100			A DEVICE-DEPENDENT UNEDITABLE
08200			REPRESENTATION JUST FOR THE XGP, OR JUST FOR
08300			512X512 TVS, OR JUST FOR THE IMLAC, OR JUST FOR
08400			THE LPT.  THE FORMATTING PROGRAM
08500			PRODUCES THIS FROM THE UNFORMATTED DOCUMENT
08600			IN PUB FASHION, ONLY BETTER (WITH COMPOSITION
08700			TECHNOLOGY INSIGHTS).  WE WILL NEED STANDARDS
08800			FOR SEVERAL TYPES OF DEVICES, BUT WE'LL START
08900			WITH THE XGP.
09000	
09100	SPROULL IS GOING TO WRITE DOWN STANDARDS FOR REPRESENTING LINE DRAWINGS.
09200	HARVEY IS WRITING DOWN STUFF ABOUT REPRESENTING MATHEMATICAL
09300	EQUATIONS AND FONT INFORMATION FILES.  I AM WRITING DOWN
09400	STUFF ABOUT PAGE LAYOUT, CROSS-REFERENCES, AND OTHER PUB STUFF.
09500	THIS ALL HAS TO DO WITH UNFORMATTED DOCUMENTS.  WE HOPE YOU ARE
09600	THINKING ABOUT FORMATTED DOCUMENTS FOR XGP AND
09700	OTHER RASTER DEVICES.  WHAT ARE YOU DOING?
09800	
09900	NOW THE LIGHTNING BOLT.  WE FIGURE WITH BRIAN
10000	HARVEY WORKING FULL-TIME UNTIL
10100	AUGUST AT S.U. AND RICH JOHNSON AT XEROX AND
10200	GREG AND HAL AT CMU, WITH A LITTLE HELP FROM ME AND JEFF RULIFSON
10300	AT XEROX WHEREVER OUR PURPOSES OVERLAP, WE SHOULD BE ABLE TO TURN OUT
10400	THE SUPEREST SYSTEM BETWEEN NOW AND THE FALL. WE MEAN TO
10500	AGREE UPON STANDARD REPRESENTATIONS AND DIVIDE THE PROGRAMMING
10600	JOB INTO PASSES (E.G., THE FORMATTER MUST FIRST PRODUCE
10700	GALLEYS, THEN PAGINATE, THEN RESOLVE CROSS-REFERENCES; THE
10800	FONT FILE MAKER MUST CREATE DEVICE-INDEPENDENT MASTERS,
10900	THEN DEVICE-DEPENDENT MASTERS, THEN FONT INFORMATION
11000	FILES).
11100	
11200	EACH PASS WILL BE PROGRAMMED IN THE MOST APPROPRIATE
11300	LANGUAGE.  WE ARE ARGUING A LOT ABOUT: SAIL, STANDARD LISP,
11400	ASSEMBLER.  E.G., FOR EQUATION LAYOUT, LISP IS BEST, BUT FOR
11500	FONT FILE PUSHING, ASSEMBLER IS PROBABLY BEST.
11600	
11700	THE COORDINATION PROBLEMS ARE OBVIOUS, BUT WITH ALL THESE
11800	TALENTED PEOPLE AND A COMMON NEED, WE THINK IT'S WORTH A
11900	TRY.  JOHN MCCARTHY HAS SAID ITS O.K. FOR HARVEY TO DO THIS,
12000	BILL ENGLISH HASN'T DECIDED ABOUT RICH JOHNSON BUT SAYS IT'S
12100	OK TO PURSUE IT AND I SHOULD MAKE RECOMMENDATIONS.  IF YOU
12200	COULD FIND OUT RICH'S FEELINGS THAT WOULD HELP.  PLUS I NEED
12300	TO KNOW YOUR REACTION TO OUR STANDARDS PROPOSALS AS WELL AS
12400	OUR JOINT IMPLEMENTATION PROPOSALS.
12500	
12600	WE ARE MEETING AGAIN TUESDAY OR WEDNESDAY TO MERGE OUR
12700	SEPARATE DOCUMENTS AND DISCUSS MORE DETAILS.  I WOULD
12800	APPRECIATE FEEDBACK OF ANY KIND BEFORE THAT.
12900	
13000	P.S.  I WONDER IF PERLIS WOULD LIKE TO GET IN ON THIS
13100	TOO.  ALTHOUGH WE MAY ALREADY HAVE PLENTY OF MANPOWER.