[PEAK] Adobe's Trellis?

Justin Ryan justizin at gmail.com
Wed Jul 18 21:20:26 EDT 2007


Whups, this bounced b/c I have changed my ML address...

On 7/18/07, Phillip J. Eby <pje at telecommunity.com> wrote:
> It seems Adobe has a C library that resembles the Trellis in some
> respects, and may be used in products such as Photoshop:
>

_is_ used in all their apps, esp Photoshop.. Adam and Eve is the crux
of Creative Suite, esp its' dual-life on win32 and Cocoa, afaik..

> http://opensource.adobe.com/group__asl__overview.html#asl_overview_adam_and_eve_adam
>
> Dubbed "Adam", it uses a custom expression language based on C, with

I have read quite a bit about this, but made no connection to Adam
when I read about Trellis.  Somehow your explanation was more
pinning..

> its own VM.  Of particular interest is something under the Goals section:
>
> http://opensource.adobe.com/group__asl__overview.html#asl_overview_goals_for_adam
>
> """Reviewing bugs across a range of Adobe products revealed that
> roughly 40% of Adobe products' bugs are "behavioral" in nature. Many
> of these bugs (4 out of 20 in the Photoshop sampling) simply could
> not have happened using Adam. Other bugs, though they may still
> occur, will be easier to find and fix or will be eliminated
> altogether by moving more of the implementation closer to the design
> process."""
>
> In other words, they're projecting a minimum 8% decrease in bugs
> overall, plus additional gains in ease of debugging or implementation.
>
> It would be interesting to see how that's worked out for them in
> practice, and also how it affects total development cost.  (Since
> hairy event-driven bugs can sometimes be more expensive to fix than
> writing entire new features.)
>

Most Definitely!  Surely there are some folks mentioned in the source
or on the f/oss website who might be interested in bragging about
their brilliant advancements.

Maybe they've published in ACM or IEEE?

--
Justin Alan Ryan
Interaction Director, VonGogo Foundation
Independent Interactivity Architect
http://www.bitmonk.net/
* : +1-415-226-1199



More information about the PEAK mailing list