From lieber Tue Jun 18 15:51:40 1991
Return-Path: <lieber>
Received: by sun375f.CCS.Northeastern.EDU (4.1/SMI-3.2+CCS-subsidiary-2.2)
	id AA28235; Tue, 18 Jun 91 15:51:37 EDT
Date: Tue, 18 Jun 91 15:51:37 EDT
From: lieber (karl lieberherr)
Message-Id: <9106181951.AA28235@sun375f.CCS.Northeastern.EDU>
To: cunxiao, huersch, nacho
Subject: propagate
Status: RO

Cun:
please can you take care of

(*task* Ask-Cun June 18 ())

-- Karl

; MAINTENANCE PLAN FOR 4.2
(*task* parse-generate
  (*task* override-cd-parse
  June 18, 91
  Override cd-parse. Parse-generate is used to debug the cd
  and cd-parse should never be changed by the user.
  Therefore parse-generate can override cd-parse.
  Discussed with Walter.
    (*task* Ask-Walter after June 28 ())
  )
 *task* propagate
  works beautifully; graphs are great help for debugging of pps
   (*task* debugging-output
     When I run propagate, I get 
     Next token, kind: 1, value: int, line number: 108
     etc.
     (*task* Ask-Cun June 18 ()))
)




From lieber Wed Jun 19 12:08:11 1991
Return-Path: <lieber>
Received: by sun375f.CCS.Northeastern.EDU (4.1/SMI-3.2+CCS-subsidiary-2.2)
	id AA29136; Wed, 19 Jun 91 12:08:09 EDT
Date: Wed, 19 Jun 91 12:08:09 EDT
From: lieber (karl lieberherr)
Message-Id: <9106191608.AA29136@sun375f.CCS.Northeastern.EDU>
To: cunxiao
Subject: maintenance
Status: RO

Cun: thank you of taking care of the messages.
Below I identified a few more maintenance tasks.

Your propagation pattern tool is great!!
Thank you for doing such a good job.

-- Karl

; To find the maintenance tasks to be done search for Ask-FIRSTNAME
; When you are done, I will change Ask- to Done-

; This way we keep track of who has done what, to give proper credit
; to the contributors.

; this file is legal with respect to
; /u2/csfaculty/lieber/project-plans/cd

; MAINTENANCE PLAN FOR 4.2
(*task* generic-parser
  I noticed that the generic parser gives several messages:
  ERROR on line:

  The current and next 10 tokens are:

  Production stack dump follows:

  Why does this happen?
  (*task* Ask-Nacho-1 June 18 ())
 *task* pre-computed-hashtable
   Cun proposed to improve make-instance and parsing
   by using a pre-computed hashtable for the classes.
   We want to keep the cost of finding the hash function small.
   We do not need a perfect hash function.
   Check Knuth, Volume 3.
   Check GNU library.
   Related to birthday problem: with 23 people in a room, the prob.
   that two have the same birthday is greater than one half.
   Need good hashing function which spreads similar names.
   We have no insertion nor deletion. Know class names in advance.
   Can randomly generate a hash function? Mehlhorn, Wegman {Ian s boss
   boss}, Carter.
   (*task* Ask-Cun-pre-computed-hashtable June 18 ())
 *task* parse-generate
  (*task* override-cd-parse
  June 18, 91
  Override cd-parse. Parse-generate is used to debug the cd
  and cd-parse should never be changed by the user.
  Therefore parse-generate can override cd-parse {saving the 
  old file for security}.
  Discussed with Walter.
    (*task* Ask-Walter-override-cd-parse after June 28 ())
  )
 *task* propagate
  works beautifully; graphs are great help for debugging of pps
   (*task* debugging-output
     When I run propagate, I get 
     Next token, kind: 1, value: int, line number: 108
     etc.
     (*task* Done-Cun-debugging-output June 18 ())
    *task* error-message
      I forgot to give a class name after *class*
      and I got a difficult to read error message
      The line number was way out of line!!
      (*task* Ask-Cun-error-message-1 June 18 ())
    *task* legal-cd
      I noticed that in the graph file generated there was
      a production: 
      Neighbors_wc  :
	      Construct_ns
		      Alternat_ns
			      *common* .
      It is important that the bars are printed.
      Also, the following is printed:
      Term : *common*.
      This might cause a problem since the propagation graph is 
      not a structurally legal cd. Should we change the cd definition?
      (*task* Ask-Cun-legal-cd-pp June 18 ())
    *task* graphical-propagation-graph
      The propagation graph given in *.graph is very useful for 
      detecting differences between propagation patterns.
      The numbers about classes help also. 
      Our eyes are good at finding differences visually and therefore
      I think that the result of xref applied to the graph should
      also be contained as a comment in *.graph.
      It is very good to keep *.graph a legal cd.
      (*task* Ask-Cun-graphical-propagation-graph-xref June 19 ())
   )
 *task* make-user
  works perfectly
    (*task* man-page
      There is no man page.
      (*task* Ask-Walter-man-page after June 28 ())
     *task* error-message
       When called in wrong directory, talks about USER.
       Change error message.
       (*task* Ask-Walter after June 28 ()))
)