launchpad-dev team mailing list archive

[Deryck Hodge <deryck.hodge@xxxxxxxxxxxxx>]

On Tue, Nov 22, 2011 at 6:23 PM, Richard Harding
<rick.harding@xxxxxxxxxxxxx> wrote:
>> On Nov 22, 2011, at 6:57 PM, Ian Booth wrote:
>> > Despite fulfilling its purpose, there's some limitations we'd like to
>> > overcome. We see these sorts of demos being used more often moving
>> > forward so coming up with a preferred way of doing them will benefit us
>> > all. One of the biggest issues for us was the restriction to static
>> > HTML, and the lack of core technologies like tales that we use in
>> > production. It would be great to have a light weight app server / web
>> > framework available which would offer more of a chance to use these
>> > technologies and to hence produce prototype code more easily transformed
>> > to something suitable for use in production once the demo has served its
>> > purpose.
>
> On the one hand I can understand, however I think one of the great
> principles of this kind of work is the chance to produce a real prototype.
> Any prototype should be trashed once complete. In this way you're
> encouraged to perhaps go out on a limb more during this stage since you're
> not going to be stuck with any decisions made except the UI interactions.
> Secondly, you don't get attached to the code so you won't spend the time to
> make it more robust and resilient since "it's going to turn into production
> code".
>

I have to agree with Rick.  The moment you start thinking about coding
more than prototyping it defeats the gain to be had from doing
prototypes first.  Why not just start coding? :)

However, I do completely understand the pain.  Doing HTML by hand is
super, super slow.  I did an HTML prototype for the custom bug
listings feature and it was an incredibly slow and tedious process.
I've done this a number of times at other companies before Canonical,
too.  The work itself is never much fun, in terms of coding.  The fun
is always in seeing the product taking shape.

I think the way to ease the pain is to allow less fully-formed
prototypes, cutting down on what you have to code up by hand.  Just
enough to show the complete interaction and no more, just enough to
understand the problems real coding will solve.

Cheers,
deryck


-- 
Deryck Hodge
https://launchpad.net/~deryck
http://www.devurandom.org/