hi all,
this is mainly addressed to people who did ports of logo/lisp like
languages.
can you describe what you did and give an estimate of how much
time did you (or your team) put into it
a) to get it running (first version) ?
b) to handle maintenance and new versions ?
c) is it done ? ( how much more time will it take ?)
most wanted answers are from people who did who a full
implementation of logo/lisp/scheme either alone or as a part of a team
on a PC enviroment (Xlisp/MSWLOGO/UCBLOGO/LOGO4)
thanks all,
eyal peleg
········@t2.technion.ac.il
········@csc.cs.technion.ac.il (Eyal Peleg) writes:
>can you describe what you did and give an estimate of how much
>time did you (or your team) put into it
> a) to get it running (first version) ?
> b) to handle maintenance and new versions ?
> c) is it done ? ( how much more time will it take ?)
Dan van Blerkom, then an undergraduate, wrote the initial version of
Berkeley Logo, doing most of the work over the summer of 1988, although he
started thinking about it during the previous semester. This version
ran only on Unix systems and didn't do graphics at all.
I worked on it during the following year, adding features and cleaning
up a few things, as a very low-priority part-time project.
Michael Katz, then a graduate student, worked on porting it to the PC
and to the Mac and adding turtle graphics. He spent a year or so on
the first release, and continued to improve it during the year after
he left Berkeley, again as a side project.
Doug Orleans, then an undergraduate, spent a year replacing the original
core evaluator with a new one based on the explicit control evaluator
for Scheme in Chapter 5 of Structure and Interpretation of Computer
Programs. This was motivated by the desire to handle tail call elimination,
which the original version didn't, because the PC port kept running out of
memory. Doug did get the explicit control evaluator to run, but didn't
entirely solve the tail call problem (which is much harder for Logo than
for Scheme); I did that over the course of the next year. The year he
spent included a semester in which he wrote no code, but just read SICP
and learned what he was trying to do.
At the same time I also moved the PC version from Turbo C to Zortech C
so that I could use their DOS extender to increase the amount of memory
available for Logo programs. This was a royal pain because of differences
in the way the two C compilers handle graphics and things like that.
Several other people, some of whom I've never met, have contributed
smaller additions and bug fixes to Berkeley Logo. Fred Gilham wrote the
X11 graphics module for Unix. George Mills has found a zillion bugs
while working on MSWLogo. Randy Sargent spent a weekend finding a
really pernicious memory leak that I'd been banging my head against
for a year. Other people have also found bugs.
Berkeley Logo is still under development. Just the other week I found a
serious bug in the array mechanism. People keep wanting things like the
ability to send the graphics screen to a printer; I'd love for someone
to volunteer to do that in a uniform way across platforms and printers!
Ports for Amiga and Atari would be nice, if someone wanted to do them.
The core evaluator is still messier than I'd like, but I doubt if I'll
ever have the time to rewrite it completely, which is what it needs.
Someone who knows more about compiler theory than I do should take
that on! I think it's about as good as I can get it by tinkering.
Still, I think the language is essentially finished. I don't anticipate
making any major changes in the language specification -- other people
are doing stuff like object-oriented versions of Logo, and they get
paid for it, whereas I have to squeeze it in.
Dan, Michael, and Doug are all brilliant programmers, but were all
learning as they went along, so I honestly don't know whether it took
them more or less time than a pro would have taken.
In article <··········@agate.berkeley.edu> ··@anarres.CS.Berkeley.EDU (Brian Harvey) writes:
>········@csc.cs.technion.ac.il (Eyal Peleg) writes:
>>can you describe what you did and give an estimate of how much
>>time did you (or your team) put into it
>> a) to get it running (first version) ?
>> b) to handle maintenance and new versions ?
>> c) is it done ? ( how much more time will it take ?)
>
>Dan van Blerkom, then an undergraduate, wrote the initial version of
>Berkeley Logo, doing most of the work over the summer of 1988, although he
>
> ...
>
>Doug Orleans, then an undergraduate, spent a year replacing the original
>core evaluator with a new one based on the explicit control evaluator
>for Scheme in Chapter 5 of Structure and Interpretation of Computer
>Programs. This was motivated by the desire to handle tail call elimination,
>which the original version didn't, because the PC port kept running out of
>memory. Doug did get the explicit control evaluator to run, but didn't
>entirely solve the tail call problem (which is much harder for Logo than
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
>for Scheme); I did that over the course of the next year. The year he
^^^^^^^^^^
Could you give us a thumbnail sketch of why this is so?
I think I know, but inquiring minds want to know for sure.
Dale Parson, Bell Labs, ····@aloft.att.com
----------------------------------------------------------------------
| "These words are too solid, they don't move fast enough |
| to catch the blur in the brain that flies by, and is gone..." |
----------------------------------------------------------------------
| Suzanne Vega |
----------------------------------------------------------------------
From: ···@paradigm.com
Subject: Re: how much time to develop system ?
Date:
Message-ID: <3506@paradigm.com>
In article <··········@discus.technion.ac.il>, ········@csc.cs.technion.ac.il (Eyal Peleg) writes:
> hi all,
> can you describe what you did and give an estimate of how much
> time did you (or your team) put into it
> a) to get it running (first version) ?
SIOD version 1.0 included Scheme READ/EVAL/PRINT and a stop-and-copy GC.
It was in a single source file of about 800 lines of code and was
implemented in one sitting of about 11 hours. At that point there was
enough to use it in some homework I assigned to my students in an
AI-programming course, EK201 at Boston University.
> b) to handle maintenance and new versions ?
I might spend about an hour or two every other month adding a feature
or encorporating some user-supplied code.
> c) is it done ? ( how much more time will it take ?)
But then I use it as a testbed for gaining experience with
new operating system and/or compiler environments.
The port to more fully ANSI C compilers, DEC C, GNU C with -Wall,
and C++ compilers certainly took more than the original 11 hours spent
in initial creation. One can easily spend hours learning the best
way to set up a makefile on a new operating system, how to create
a shared library, how to best use the debugger, etc.
-gjc
Greetings.
>Berkeley Logo is still under development. Just the other week I found a
>serious bug in the array mechanism. People keep wanting things like the
>ability to send the graphics screen to a printer; I'd love for someone
>to volunteer to do that in a uniform way across platforms and printers!
>Ports for Amiga and Atari would be nice, if someone wanted to do them.
I read your comments about the development of Berkeley Logo with interest.
I am constantly amazed at how nasty bugs can lurk unseen for quite some
time even after you think you've tested the heck out of something; and yet
other bugs which you can "bang your head against" as you say, for months
and then one day the light dawns.
I have spent the past 2 years or so developing an Amiga BASIC compiler. It
even has turtle graphics primitives. :-) I also wrote a rather cut down
Logo interpreter a few years ago for the Amiga, and have thought of
creating a full version. Perhaps Berkeley Logo for the Amiga would be
something I could work on?
Could you give me an estimate on what areas would need Amiga-specific
porting and how large you think the job might be? I'm still working on my
compiler and expect to be doing so for some time, but I might also be able
to pick up a project such as this.
Anyway, contact me via e-mail if you think I might be able to help.
Regards,
David Benn
--
······@appcomp.utas.edu.au - David Benn. University of Tasmania at Launceston.
It seemed so simple when one was young, and new ideas were
mentioned not to grow red in the face and gobble. (Logan Pearsall Smith)