Re: OT: An Udanax-Gold implimentation in Dolphin-ST

----- Original Message -----
From: "David G Jones" <david_jones@xxxxxxxxxxxxxxxxxx>
To: <squeak-dev@xxxxxxxxxxxxxxxxxxxxxxxxxx>
Sent: Wednesday, September 25, 2002 8:27 PM
Subject: Re: OT: An Udanax-Gold implimentation in Dolphin-ST


> "Les Tyrrell" <tyrrell@xxxxxxxxxxxxxx> writes:
> > Once upon a time I had similar ambitions...
> >
> > http://griffin.canis.uiuc.edu:8080/Xanadu
> >
> > I am *really* impressed at the progress you've made from scratch- the
source
> > code I have from Xanadu remains very baffling to me.
>
> Thanks for the link, it is the first time I have seen your work - I
> somehow missed it on my numerous searches through the net looking for
> bread crumbs.

For quite a while I had problems on my wiki server that kept it from bringing
this one back online, so that is not too surprising.


> I initially tried auto-translation to Java - spurred on by the type
> info, then retreated back to a similar strategy to yours of loading in
> tweaked versions into a current Smalltalk. I managed to get some of
> the tests to partially run. Unfortunately the tests aren't self
> checking, and seem to require diff'ing against the results of a known
> good run.
>
> Just like you my efforts faltered while working with the original
> code; primarily due to its large size, generality, lack of
> documentation and mysterious naming. It's also unclear of what was
> left to do, and open bugs, at the time the original development
> ceased.

When this project wound down for me I was coming to the conclusion that these
sources did not actually represent a working system at all- instead, my best
interpretation of this was that they were the sources of some sort of
macro-assembler for the CXX language which provided the actual implementation.
As best as I understand this, the Xanadu team had licensing concerns regarding
ObjectWorks, so they chose to develop using ObjectWorks but deploy by
translating to their own C-like language with object extensions.  The
nomenclature is rather tough to follow, but I've seen better explanations more
recently about these things.  Chances are you've found them, but I think I
have links to those sites on my wiki.


> Your work seems to be a lot more organized than my importing efforts,
> and I'm impressed that you managed to get some of the missing classes
> from the XOC people.
>
> I am dipping into the original source for my Abora app, but the best
> headway to understanding Udanax-Gold seems to be a paper on a
> simplified form of the Ent, and the first couple of chapters of the
> Xanadu Information Architecture docs. I think both are referenced from
> the Abora readme files.

That sounds like the best strategy to me- grokking the UG sources is a pretty
tough route to go, especially as I believe that these sources are not what
they may at first seem to be... that is, UG was not a Smalltalk application,
rather, Smalltalk was used as some sort of assesmbler for the Xanadu system.
I really have a hard time believing that there is a functional system in the
pure Smalltalk code.

Good luck - someday I hope to pick up on these projects again, but time has
been very limited lately.

- les