Re: "everything is a file" and beyond

From: Paul Hepworth <paul1+vsta_at_nospam.org>
Date: Thu Feb 26 2004 - 15:24:35 PST

<vandys@vsta.org> wrote:
> This implicitly accepts that the goal is to be expressively comparable
> to XML. I don't see why that should be accepted as a goal? Both Plan9
> and VSTa have become fully functional systems using only filesystem
> semantics. Remember, the goal of VSTa is not to do everything everyone
> else can do, but to do just those things which make sense, and do them
> in a way which minimizes the number of architectural constructs and thus
> (hopefully) maximizes implementation elegance.

No, it doesn't imply any goal at all. It merely suggests how one could
extend filesystem-like syntax to support a significant additional feature
that XML has: order-preservation of children. No other goal, no hidden
agenda, no gripe about the current system. I was just bringing up an idea,
not saying that this idea automatically should be implemented or is
automatically the best thing for VSTa. I've long considered XML to be 90%
hype, 10% interesting data format. I do like some aspects a lot, though,
especially the way elements nest. Where data nesting isn't needed, good old
comma-separated value or key-value pairs work just fine. However, nesting
(with order-preservation of children) has genuine utility for some
appliations, and I was inviting discussion on this. Discussing something
doesn't imply that it should immediately be someone else's goal.
Indeed, when I started this thread, I invited opinions about whether this
was a good fit for VSTa with the following:
    Opinions?
    What sort of XML messages would be needed/desired? (In addition to the
    messages used for virtual filesystems, we'd need to add general
attribute
    read/write and order-preserving container support (iterators,
    order-preserving insert/delete))
    Is VSTa the best platform for such an experiment?

It was certainly not my intention to offend anyone. I guess you've got some
harsh feelings toward XML from your past project. :-)

Now, it would indeed be fair to say that this implicitly accepts that the
goal of my idea of filesystem-like handling of XML is to make parsing XML
less cumbersome. I think it's less cumbersome than common XML libraries such
as SAX and much more intuitive to anyone familiar with filesystems. For
accessing specific elements, it's very intuitive. For iterating over
elements, it's not extremely different from iterating over multiple files.

Anyway... my current interest is more in the UI area (widgets and such),
which can be done just fine with or without order-preserved children (the
size and position attributes of the widgets define the "visual" order); with
stat/wstat, I have what I need. (I even called it "perfect" in an earlier
post.)

Paul
Received on Thu, 26 Feb 2004 16:24:35 -0700

This archive was generated by hypermail 2.1.8 : Tue Sep 26 2006 - 09:03:11 PDT