Self-interest May 1991

self-interest@lists.selflanguage.org

10 participants
11 discussions

Re: Binary data and set. & other things
by Parag Patel 31 May '91

31 May '91

I'm interested in your OO extensions and would like to be on your mailing list. I'm "Parag Patel" <parag(a)sde.hp.com>. Thanks! Just for fun, a while ago I had built a primitive class mechanism for TCL entirely written in TCL! Here it is to add to your "truly deranged" archives. These files work with the Tcl 4.0 release as I use a lot of the 4.0 functions like incr, clength, split, etc. It could be made to work with Tcl 5.0, but I haven't had the time to hack it yet. Sorry! Just source "class.eg" for the example code of how classes work. (I use a Tcl proc named "@" as the basic "send-object-message" routine.) -- Parag ---- Cut Here and feed the following to sh ---- #!/bin/sh # This is a shell archive (produced by shar 3.49) # To extract the files from this archive, save it to a file, remove # everything above the "!/bin/sh" line above, and type "sh file_name". # # made 05/31/1991 18:10 UTC by parag@hpsdeb # Source directory /users/parag/lib/tcl # # existing files will NOT be overwritten unless -c is specified # # This shar contains: # length mode name # ------ ---------- ------------------------------------------ # 529 -rw-r--r-- class.eg # 3345 -rw-r--r-- class.tcl # # ============= class.eg ============== if test -f 'class.eg' -a X"$1" != X"-c"; then echo 'x - skipping class.eg (File already exists)' else sed 's/^X//' << 'SHAR_EOF' > 'class.eg' && source class.tcl X # define class "Foo" with construct, destructor, and message "Get" class Foo {value string} cproc Foo Foo {} { echo Constructed!; cset string "<default>" } cproc Foo ~Foo {} { echo Destructed! } cproc Foo Get {} { return [cset string] } X # define class Bar derived from Foo with an addition message "Set" class Bar:Foo {} cproc Bar Set {s} { cset string $s } X # create a new Bar object - both Set and Get should be available set foo [new Bar] echo [@ $foo Get] @ $foo Set NewValue echo [@ $foo Get] delete $foo SHAR_EOF chmod 0644 class.eg || echo 'restore of class.eg failed' Wc_c="`wc -c < 'class.eg'`" test 529 -eq "$Wc_c" || echo 'class.eg: original size 529, current size' "$Wc_c" fi # ============= class.tcl ============== if test -f 'class.tcl' -a X"$1" != X"-c"; then echo 'x - skipping class.tcl (File already exists)' else sed 's/^X//' << 'SHAR_EOF' > 'class.tcl' && # Prototype of Class extensions to TCL # by Parag Patel X set classlist "" set inheritlist "" X proc class {class vars} { X global classlist inheritlist X X # get the inheritance info, if any X set inh [split : $class] X set class [index $inh 0] X set inherit "" X if {[length $inh] == 2} { set inherit [index $inh 1] } { X if {[length $inh] != 1} { X error "illegal inheritance syntax for class $class" X } X } X X # each elt in inheritlist in the name of the parent class X # - the key is the child class which is derived from the parent X set inheritlist($class) $inherit X X # each elt in classlist is a list of its class vars X set classlist($class) $vars } X proc cproc {class mesg arglist body} { X # add "this" to the arglist and define a proc "class.mesg" X set arglist "this $arglist" X proc $class.$mesg "$arglist" $body } X proc cset {var args} { X set this [uplevel {list $this}] X global $this X if {[clength "$args"]} { X set ${this}($var) $args X } { X return [set ${this}($var)] X } } X set objectid 0 proc new {class args} { X global classlist inheritlist objectid X X # create a new object handle X set this "@Object_$objectid" X incr objectid X X # the handle is also an array indexed by its class var names X # - thus the values for these array elts are unique for each object! X global $this X X # initialize all the class vars to empty X # - we need to walk the inheritance tree to get ALL the vars X set clist "" X set inhlist "" X set varlist "" X for {set c $class} {[clength "$c"] > 0} {set c [set inheritlist($c)]} { X foreach {v} [set classlist($c)] { X set ${this}($v) "" X set varlist "$v $varlist" X } X set clist " $c$clist" X set inhlist "$inhlist $c" X } X set clist "$c$clist" X X # these are just to make message dispatching easier in "@" below, and X # so that delete can call the right destructors - at the very least, X # "class.name" must always be available to extract everything else X set ${this}(class.name) $class X set ${this}(class.variables) "$varlist" X set ${this}(class.inheritance) "$inhlist" X X # invoke the constructors in the correct order, deepest parent 1st X foreach {c} $clist { X if {[clength "[info procs $c.$c]"]} { X if {[clength "$args"]} { @ $this $c $args } { @ $this $c } X } X } X X # and return our new object's handle X return $this } X proc delete {this} { X global $this X X # sanity check - do not allow deleting a deleted object X if {![info exists ${this}(class.inheritance)]} { X error "no object $this exists to delete" X } X X # call the destructors in reverse order of constructors, child 1st X foreach {c} [set ${this}(class.inheritance)] { X if {[clength "[info procs $c.~$c]"]} { @ $this ~$c } X } X X unset $this } X proc @ {this mesg args} { X global classlist inheritlist $this X X # find a message proc following the inheritance chain X set procname "" X foreach {c} [set ${this}(class.inheritance)] { X set p $c.$mesg X if {[clength "[info procs $p]"]} { set procname $p; break } X } X if {![clength "$procname"]} { X error "No message $mesg defined for class [set ${this}(class.name)]" X } X X # invoke the message proc adding "this" to the arguments X if {[clength "$args"]} { X return [$procname $this $args] X } { X return [$procname $this] X } } SHAR_EOF chmod 0644 class.tcl || echo 'restore of class.tcl failed' Wc_c="`wc -c < 'class.tcl'`" test 3345 -eq "$Wc_c" || echo 'class.tcl: original size 3345, current size' "$Wc_c" fi exit 0

1 0

Re: Binary data and set. & other things
by Sean.Levy＠cs.cmu.edu 31 May '91

31 May '91

I know that this might be much more than people were asking for, but I have implemented an object system on top of Tcl. It most closely resembles Self in its treatment of objects (prototype-based, multiple inheritance, etc). The C library that implements the basic object functionality is very small; methods are, of course, implemented as hunks o Tcl code that are invoked in an interesting way :-) We are still in the process of testing things out, mainly by building wrapper "classes" (traits in Selfese) around the Tk widgets, writing a graphical browser on top of these, and a couple other things. Preliminary testing seems to indicate that the overhead is very small when compared to what you get (never having to say "global" again, among other things). I use the same method that Tk uses to give the world access to "binary" data (e.g. objects): Tcl commands. I think this is a pretty good method to use in general for this sort of problem in Tcl. Bind clientData to whatever strange thing (array of floats, etc) you want, have one function in C that all such data object commands get bound to, and you're pretty much done. I intend to use the C side of my object system to implement things like Hash tables, arrays, etc. in Tcl; in this case, the Tcl side is already done, I just have to write some methods in C to dick with the kinds of objects at hand. If anyone is interested in the Basic Object System, please drop me a line and I'll add you to a list. It should be beta-testing in a week or two. Cheers, Sean Levy <snl(a)cs.cmu.edu> nDim Project, Engineering Design Research Center Carnegie Mellon University, Pittsburgh, PA

1 0

Please add me to your mailing list
by Thomas Lindgren 28 May '91

28 May '91

Please add me to your mailing list. Thomas -------------------------------------------------------------------------- Thomas Lindgren thomasl(a)emil.csd.uu.se Computing Science Dept., Thomas.Lindgren(a)csd.uu.se Uppsala University -----------------------------\\\\-----------------------------------------

1 0

Re: Is Printing Meta?
by David.Ungar＠Eng.Sun.COM 25 May '91

25 May '91

As far as I can tell, there are (at least) two kinds of printing. Print an object as an object (just see its slots), and print it as whatever it represents (e.g. 3@4). The former is meta, the latter isnt. Anyway, that's my best story about it for now/ Dave

1 0

Re: Is printing meta?
by xanadu!hibbert＠uunet.UU.NET 25 May '91

25 May '91

Those are good points, John. Besides making it more obvious how to find the code, the reason I think printing belongs in the object is that it necessarily involves looking in the guts at the detailed state. (There are times when you want to print different amounts of information depending on who's asking, but I think the way to detect that is by using different messages to ask.) If you have mirrors (really should be lenses or windows) then they have to ask the object for the information anyway, and so you've reduced to the previous problem anyway. The object itself is the only thing that can really know enough to construct a reasonable representation. I didn't understand what you meant in your last line: >> mirrors are a lot like views, which >> are considered bad (at least in the UI) I thought that the idea of Views as in MVC was still considered to be a good one. Having objects that are in charge of knowing how to display others that respond to a given protocol (or can be made to by some mediating object) is a real win. Are you talking about some other facet of the idea of views, or are there more recent evaluations of the general idea that I haven't seen yet? Chris

1 0

Re: Is Printing Meta?
by jmaloney＠cs.washington.edu 24 May '91

24 May '91

(apologies to those who have seen this twice; i sent to the wrong list the first time...) Gee, I dunno. I always liked having objects in charge of printing themselves. As a programmer, I always knew where to look for the printOn: method and writing "foo printString" is more concise than "foo mirror printString", or whatever it would be. Finally, "meta" anything often confuses beginning programmers because they think it must be more profound than it really is. Maybe I don't have the context to understand the philosphical issue here but it seems to me that mirrors are a lot like views, which are considered bad (at least in the UI)... -- John

1 0

Is printing meta?
by Randall.Smith＠Eng.Sun.COM 23 May '91

23 May '91

Should we really send "printString" to an object? Perhaps it is something one really should ask of a mirror on an object. After all, a name is a representation of an object, so it's kind of meta (mirror or meta = about).

1 0

editor
by R James Noble 08 May '91

08 May '91

The emacs editor sounds interesting - I presume it's written using the SunView UI? - one of the messages I never got around to posting was to solicit ideas on a device (window-system) independant output model for Self (so that me using X didn't lose to much). Oh well. So why not package it up, (or even just dump a snapshot on contrib ?). I would like more "application" self code to look at. james R James Noble, Graduate Student, Computer Science Department Victoria University, Box 600, Wellington 1, New Zealand kjx(a)comp.vuw.ac.nz

1 0

Emacs like editor ...
by eat＠scs.carleton.ca 07 May '91

07 May '91

Hello Self world, I have just finished (actually 2 weeks ago, but I just got time to write this message today) writing an editor that is somewhat Emacs like in the graphical UI of Self. It is nothing sepctacular right now, but it can read files and write them back out after editing. Currently only the following functions are supported (I think !!!): - ctrl-n: cursor down - ctrl-p: cursor up - ctrl-b: cursor left - ctrl-f: cursor right - ctrl-v: page down - esc-v: page up - ctrl-d: delete char over cursor - backspace: ... - ctrl-x-s: save file - ctrl-k: kill to end of line - ctrl-y: yank out what was killed with ctrl-k I am about to go on and work on some more features: - lines longer than one screen - currently lines that are wider than the editor is are not supported correctly, the current hack is that you can open up the editor as big as you wish - save as - search - ctrl-u: repeat the following sequence - status bar, much like the Emacs one What I was wondering was, should I spend any time to try and package what I have now and release it to you guys, with the full understanding that the code is not very readable (this is my first Self project) and that there may be bugs. If anybody would like what I have now, then send me some E-mail and I'll release it to the contrib if I can get enough responses to warrant the time to package ... :) Eric Arseneau :) eat(a)scs.carleton.ca Danny Epstein dje(a)scs.carleton.ca

1 0

name spaces
by vme131!lsiserv1!jecel＠uunet.UU.NET 03 May '91

03 May '91

Tree or graph shaped name spaces are not always easy to organize. The lobby and name inferencer in Self are nice, but here is an alternative I am trying : set intersection. First we need a method to calculate the intersection of two sets. This is not a very good one, but it works: traits set _DefineSlots: ( | $ aSet = ( | inter | inter: set copy. do: [ | :e | ( aSet includes: e ) ifTrue: [ inter add: e ] ]. inter size == 0 ifTrue: [ ^ nil ]. inter size == 1 ifTrue: [ inter do: [ | :e | ^ e ] ]. ^ inter. ). | ) Now we make some sets accessible from the lobby : _DefineSlots: ( | nameSets* = ( | big = set copy. small = set copy. yellow = set copy. red = set copy. car = set copy. fruit = set copy. | ). | ) Place some objects in those sets : big add: 1 yellow add: 1 car add: 1 small add: 2 yellow add: 2 car add: 2 big add: 3 red add: 3 car add: 3 small add: 4 red add: 4 car add: 4 big add: 5 yellow add: 5 fruit add: 5 small add: 6 yellow add: 6 fruit add: 6 big add: 7 red add: 7 fruit add: 7 Can you guess what the following give you ? red red$car big$red$car big$fruit fruit How can I get this ( useful, in my opinion ) effect with trees or graphs? What are the disadvanteges of this method ( besides performance ) ? - Jecel

1 0

2024

2023

2022

2021

2020

2019

2018

2017

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

1996

1995

1994

1993

1992

1991

1990

Self-interest May 1991