Comments on: Screencasting and scripting

By: Michael Feldstein

Michael Feldstein — Fri, 05 Oct 2007 21:57:56 +0000

“Are those tools semantic in the sense I mean here, and not just mouse-click and keystroke based?”

See this demo:

http://www.rocketools.com/netsuite/SalesRepAnnotated/NetSuiteSalesRep2.htm

Epiplex guides are even more impressive in some respects because they do work with desktop as well as browser-based UIs.

By: engtech

engtech — Sat, 29 Sep 2007 07:48:03 +0000

Here’s a post by one of the guys who developed the UI testing software that was acquired by Rational (not sure if that’s the same Rational Robot as comment #1)

http://smoothspan.wordpress.com/2007/09/09/using-mindmaps-to-explore-user-interaction-and-create-qa-test-plans/

By: Jon Udell

Jon Udell — Thu, 27 Sep 2007 20:15:22 +0000

“I think a lot of the value in an expert’s screencast is capturing missteps and grumblings. What the expert has, and the novice doesn’t, is a sense of the limitations of the tool.”

That’s a great point. I would have edited out some of the missteps and grumblings that you left in your NY Times screencast. But it’s clearly better with them left in, for exactly the reason you say.

OTOH the 11 minutes I extracted from my 50-minute session with the Resolver folks, which elides missteps as well as less-than-compelling feature demonstrations, is exactly right for something that’s so new and experimental. There will be plenty of time later to illuminate the negative space, but for now the positive space — which is large and uncharted — wants to be lit up.

“Maybe these semantic tools need to be capturing my blood pressure in real time?”

Dunno about that but it would be, no joke, a terrific enhancement for tools that seek to capture and evaluate users’ experiences with software, e.g.:

http://weblog.infoworld.com/udell/2004/06/06.html
http://www.infoworld.com/article/04/06/04/23FEuser_1.html

By: Chris Gemignani

Chris Gemignani — Thu, 27 Sep 2007 18:14:46 +0000

To take matters a little off your intended target, I was struck by Jim’s thought: “If you had a script that could drive an application through all of the things shown in a screencast, you wouldn’t need the screencast.”. A very programmery thought, indeed. Theoretically, the two are semantically identical and can be substituted with no loss of generality, etc., etc.

I think a lot of the value in an expert’s screencast is capturing missteps and grumblings. What the expert has, and the novice doesn’t, is a sense of the limitations of the tool. Conventional training covers the positive space of what a tool can accomplish, without illuminating the negative space. Little places where you get stuck and grumble–e.g. my bitching about font handling on Windows–may convey as much meaning to the audience as the smooth parts.

I can convey this meaning with my voice. Maybe these semantic tools need to be capturing my blood pressure in real time? ;-)

By: Jon Udell

Jon Udell — Thu, 27 Sep 2007 01:15:14 +0000

“http://services.alphaworks.ibm.com/coscripter/browse/about”

Actually CoScripter is where this whole discussion started, here:

http://blog.jonudell.net/2007/09/10/automation-and-accessibility/

and before that here:

http://blog.jonudell.net/2007/09/06/the-social-scripting-continuum/

The follow-on question is: How to achieve these effects across application styles (desktop / RIA / browser)?

“IBM’s Rational Robot or even Mercury’s WinRunner”

Are those tools semantic in the sense I mean here, and not just mouse-click and keystroke based?

By: slimamamou

slimamamou — Wed, 26 Sep 2007 16:50:42 +0000

here it is :

http://services.alphaworks.ibm.com/coscripter/browse/about

it’s a browser plugin, with wiki. you can script, share, edit and automate nearly anything you do in web. (very useful also for testing)

By: Michael Feldstein

Michael Feldstein — Wed, 26 Sep 2007 13:45:52 +0000

A product called Epiplex from a company called Epiance (http://www.epiance.com/) has been able to do what you are suggesting for years. They used to have full UI-based automation that they essentially promoted as a cross-application macro tool but found it to be fragile enough (e.g., it tends to break when screens are updated) that they appear to have stepped back from that particular application of the technology. (It’s been a while since I’ve had contact with the company, so I’m going by what they have on their web site.) However, the product still does support interactive cue cards that walks a person through the steps of a process and monitors their progress using the same basic UI automation technology. The guy who was the real genius behind the product design, Gary Dickelman, is no longer with Epiance but runs his own company, EPSSCentral (http://www.epsscentral.com). You’ll want to talk to him; he knows more about what’s been done with this stuff than anyone.

In the same vein as epiplex, if you’re content to work only with HTML-based UIs, is ActiveGuide by Rocketools (http://www.rocketools.com/).

At any rate, the idea that you are describing is good, and it works.

By: David Ing

David Ing — Wed, 26 Sep 2007 13:32:30 +0000

Jon,

It sounds a little like you’re looking for something like IBM’s Rational Robot or even Mercury’s WinRunner. From the different aspect of testing they need to try solve a similar problem of recording/playing back direct UI-based interaction for the development/testing cycle.

Maybe a light-weight version of those kind of tools will emerge? Goodness know the Behavior/Test Driven and Unit Testing styles would benefit from better tools like this on the development side…