Re: On working with existing systems

From: ncarroll (
Date: Sun Aug 27 2000 - 11:53:23 PDT

Eric Armstrong wrote:
> ncarroll wrote:
> >
> > And in 30 years I would rather look back and say "We did it --
> > and changed the world", than
> >
> > ... "We had the best system." (I coulda been a contenda....)
> >
> Couldn't agree more. Question is, what's the best strategy for
> doing that?

Thank you. Yep, that's a question. I feel some embarrassment at
butting into architecture issues with pesky marketing thoughts,
and only do so because I have too many memories of being handed
a finished product that was technically great, but off-target
for the market. Anyhoe, more below ....
> > Can you guys come up with an architecture that accommodates
> > older email via a workaround "module"? Something -- perhaps
> > less than perfect -- which allows older email clients to
> > plug in, but doesn't trash the basic architecture of OHS?
> >
> Dunno. Lee really seems to feel its not much of a problem.
> I tend to be less sanguine, but I could be wrong. His vision
> was based on working from version-controlled files that
> would enable change-detection. That's possible, but everytime
> I try to visualize such a system, I see a lot of problems.

Over my head. I guess I was thinking of some slap-dash add-on,
which would give partial functionality to 3/4/5 email clients.
Initially bundle it as a whole, then one happy day the web site
offers two versions, lean and mean XML Version (with the
module gone), and the Older Version ... thankfully fading
into history.

> Without version control, the issue of identifying what
> part(s) of a reply are copied from a prior message is
> problematic -- for example, I tend to reformat lines to get
> good line breaks and fix typos in the original text! That
> will screw the system up big time, I think.

Yeah, courteous email practices can be a problem.

While not a full solution, I'd note that email etiquette is
fading fast, which may suit you coders just fine. Where
old hands break up emails and answer point by point,
wielding [snip] occasionally -- most people simple shove
the sender's copy down and type their response above.

> The vision I have of the result looks like a half-baked thing
> that will work well some of the time, but look weird a lot
> of the time, too.
> I'd like to be wrong about it all, but at the moment that's
> asking me to set aside a lot of experience that's telling me
> how things are going to go. I'll be having additional
> conversations on the architecture in the future. We'll see...




Nicholas Carroll Email: Alternate: ______________________________________________________

This archive was generated by hypermail 2.0.0 : Tue Aug 21 2001 - 17:57:53 PDT