Udanax ? (was Persist or not persist : comment)

btanksley@hifn.com btanksley@hifn.com
Tue, 26 Oct 1999 14:47:02 -0700


> -----Original Message-----
> From: Jecel Assumpcao Jr [mailto:jecel@lsi.usp.br]
> Subject: Re: Udanax ? (was Persist or not persist : comment)

> shapj@us.ibm.com wrote:

> > 4. A design that really hadn't been thought out well with 
> > regard to common cases
> > and/or simplification and/or I/O management.
> [...]
> > The Xanadudes were quite brilliant architects. Problem (4), 
> > however, is serious,
> > and I think that any system built from either of the 
> > original Xanadu code bases
> > will fail. I am hopeful that udanax will eventually 
> > succeed, but I don't think
> > that they will do so from the current code base.

> Thank you very much for sharing this insight, which I think is
> particularly relevant to Tunes (and even more to my own project,
> since I have worked on it for so many years without releasing
> anything yet).

Agreed.  It would be ironic if one of us in the Tunes camp were to criticise
Xanadu ;-).

> Some folks over at the Squeak camp are excited that the second
> version was written in Smalltalk and are hoping to port it and
> use it for source code management/versioning. I think the
> architecture at least (if not the source code itself, which I
> haven't looked at) might work very nicely for that.

That's my opinion as well.  I also believe that the source code is more a
liability than an asset -- the main thing that's good about the open
sourcing is the ideas and the patent protection.

> I have tried to learn all I could about Xanadu since it was
> described in Byte magazine some 11 years ago, 

Cool.  I've been doing the same thing since it got released.

> but decided that
> while it sounded exactly like what I needed, it would probably
> not perform very well as a persistent object system (too many
> links for too little content).

I'm actually thinking that it might work well as a file system (well, no
files, but you get the idea).  I suspect -- without proof -- that people
will be willing to buy more storage space in return for the benefits.  I've
talked to several casual computer users about some of the use cases I
envision, and their eyes light up -- I think it's solid.

> BTW, some pages hint that an idea called "Ent" is key to
> understading the architecture of the second Xanadu version, but
> I could not find anything at all about it online.

Ent is the data structure.  You have to read the code to figure it out.

I also found that reading ALL the related pages was almost essential -- for
example, read all the projects linked to from www.xanadu.co.au.  Examine
ZigZag and whatever they're calling the version control system today.  IMO,
none of these systems is interesting in and of themselves (the UI for ZigZag
is a disaster); but they all help point toward what Xanadu is supposed to
be, and how it's supposed to do it.

If you gain any real understanding, perhaps we should work together to get
an implementation.  It may not have anything to do with Tunes, but it's
certainly more useful than doing nothing.

> -- Jecel

-Billy