Transcript **To bring - EUSES Consortium

End User Software Engineering
CHI’2008 Special Interest Group Meeting
Tue, April 8, 2008, 9:00 AM - 10:30 AM
Brad Myers, Margaret Burnett,
Mary Beth Rosson, Andrew Ko, Alan Blackwell
-1-
Definitions
• (Note: may be controversial—defer
discussion until later)
• Program
– ‘‘A set of statements that can be submitted as a
unit to some computer system and used to direct
the behavior of that system’’
– Oxford Dictionary of Computing
• Programming
– ‘‘The process of transforming a mental plan of
desired actions for a computer into a
representation that can be understood by the
computer’’
– Jean-Michel Hoc and Anh Nguyen-Xuan
End-User Software Engineering SIG at CHI’2008
-2-
Definitions, cont.
• Professional Programmer
– Someone whose primary job function is to write
or maintain software
– Typically have significant training in programming
(e.g., BS in CS)
• Novice Programmer
– Someone who is learning to be a professional
programmer
End-User Software Engineering SIG at CHI’2008
-3-
Definitions, cont.
• End-User Programmer (EUP)
– People who write programs, but not as their
primary job function
– Instead, they must write programs in support of
achieving their main goal, which is something else
– Covers a wide range of programming expertise
• Business executives and secretaries
• Physicists
End-User Software Engineering SIG at CHI’2008
-4-
Examples of EUP
• Accounting (spreadsheets)
• Analysts using MatLab
• Creating a web page
• Recording Macros in Word
• Automating office tasks
• Business software (SAP programming)
• “Programming” VCRs, Microwaves
• Scientific research
• Authoring educational software
• Creating email filters
• Musicians configuring synthesizers
• Entertainment (e.g., behaviors in The Sims)
• Web 2.0: Mashups, end-user created content
End-User Software Engineering SIG at CHI’2008
-5-
Other Names
• Also called “End User Development” (EUD)
• Some “Domain-Specific Languages” (DSL)
– Often created for end-user programmers
• Visual (Graphical) Programs
– Sometimes created for EUP
• “Scripting” languages, “Macros”
• Rapid Application Development (RAD)
End-User Software Engineering SIG at CHI’2008
-6-
Definitions, cont.
• End-User Software Engineering (EUSE)
– End-User Programming with the addition of
systematic and disciplined activities that address
quality issues (“attributes”)
• E.g., analyzing code, understanding unfamiliar code,
testing code, checking code against a model,
sharing code with coworkers, maintaining code, and
deploying code, to name a few
End-User Software Engineering SIG at CHI’2008
-7-
End User Programmers
• A very large group
• In 2012: — Scaffidi, Shaw and Myers 2005
– 90 million computer users at work in US
– 55 million will use spreadsheets or databases at work (and
therefore may potentially program)
– 13 million will describe themselves as programmers
– 3 million professional programmers
• We should make better tools for all of these people!
100,000,000
80,000,000
60,000,000
40,000,000
20,000,000
0
Users
Spreadsheets Self-Described
Professional
and
DBs
Programmers
Programmers
End-User Software
Engineering
SIG at CHI’2008
-8-
Evidence of Need for EUSE
• Why need Software Engineering for EUP?
• Lots of errors attributed to End-User Programming
of spreadsheets:
– Columbia Housing Authority admitted to overpaying by
$118,387 due to a spreadsheet data-entry error (February
22, 2006)
– New York Times, Oct 30th, 2003 - $1.2 Billion Spreadsheet
Error at Fannie Mae
– TransAlta Corp. took $24 million charge to earnings due to
cut-and-paste error in an Excel spreadsheet (June 3rd,
2003)
– Auditor, major accounting firm:
“...in 6 years work, checking literally hundreds of businesscritical models, ... my team have never failed to find errors.”
– …. (many more!)
– See http://eusesconsortium.org/euperrors/
End-User Software Engineering SIG at CHI’2008
-9-
Consequences, 2
• Also, errors in:
– Web pages
– Email filtering rules
– … (what else? Need more examples!)
• From the WEUSE II workshop at CHI’2006:
– Clinical customization package used by medical personnel
reports the need for better reuse and debugging support
– SysAdmins need better testability of database and other sorts
of scripts
– Issues with reuse of MATLAB applications
End-User Software Engineering SIG at CHI’2008
- 10 -
So What is Happening?
• Growing group of researchers in EUSE
• The EUSES Consortium
– End Users Shaping Effective Software
– http://eusesconsortium.org
– 7 sites, 13 researchers + ~15 students
• EUDNet
– http://giove.isti.cnr.it/eud-net.htm
– European Commission
– 2002-3
– 16 partners
End-User Software Engineering SIG at CHI’2008
- 11 -
Addressing the EUSE Problem
• A multi-(sub)disciplinary problem needing:
– Software engineering research
– Programming language research
– Education research
– End-user programming research
– HCI research of all types
• This is a big job, needs a whole community!
– Hence, this SIG, to encourage interested people
to work together with us and with each other
– Also, connect researchers and EUP vendors
End-User Software Engineering SIG at CHI’2008
- 12 -
Many Sessions at CHI’08
• This kind of work is represented in many sessions
– Tues, 9:00-10:30am (Now ): “Aesthetics, Awareness,
and Sketching” paper session
– Tues, 11:30-1pm: “Beyond End-User Programming”
paper session
– Tues, 4:30-6pm: “Invited Session: The Next Challenge:
from Easy-to-Use to Easy-to-Develop. Are You Ready?”
– Wed, 4:30-6pm: “Activity-Based Prototyping and
Software” paper session
– Thurs, 2:30-4pm: “End-Users Sharing and Tailoring
Software” papers session
End-User Software Engineering SIG at CHI’2008
- 13 -
Previous Meetings on This Topic
• Dagstuhl Conference on End-User Software
Engineering (WEUSE-III) on Feb, 2007
• 2nd Workshop on End-User Software
Engineering (WEUSE-II) at CHI 2006
• 1st Workshop on End-User Software
Engineering (WEUSE-I) at ICSE 2005
• CHI 2007, 2005, 2004 SIGs on End Users
Creating Effective Software
• Many proceedings and notes available off of
EUSES web page:
http://eusesconsortium.org
End-User Software Engineering SIG at CHI’2008
- 14 -
Future Meetings
• Fourth Workshop on End-User Software Engineering (WEUSE IV)
at ICSE'08 on May 12, 2008 in Leipzig, Germany
– http://eusesconsortium.org/weuse/
• Cooperative and Human Aspects of Software Engineering
(CHASE) at ICSE’2008 on May 13, 2008 in Leipzig, Germany
– http://softwareresearch.ca/seg/CHASE/
• ICSE’2008, May 10-18, 2008 in Leipzig, Germany
• PPIG’2008, Sept 10-12, 2008, Lancaster, UK
– http://www.comp.lancs.ac.uk/ppig08
• 2008 IEEE Symposium on Visual Languages and Human-Centric
Computing (VL/HCC 2008), Sept. 16-20, 2008, Herrsching am
Ammersee, Germany
• Intelligent User Interfaces (IUI’2009), January ??, 2009
• Second Symposium on EUD, March 2-4, 2009, Siegen, Germany
– http://www.eud2009.org
• CHI’2009
End-User Software Engineering SIG at CHI’2008
- 15 -
Ways to Get Involved
• Attend upcoming events.
• Get newsletter on end-user
programming/software engineering:
– (Low-volume). Sign-up sheet going around.
• Tell us today what you’re working on.
• Help shape future events (starting with
today’s discussion).
End-User Software Engineering SIG at CHI’2008
- 16 -
Rest of This SIG
• Alan Blackwell: short introduction to EUDNet
• Andy Ko: overview of dimensions of organizing
EUP and EUSE systems and concerns
• What companies are working on that is
relevant to EUP and EUSE
• Break up into groups by Software Engineering
Issue
• Reconvene and report (15 min – 3 min each)
End-User Software Engineering SIG at CHI’2008
- 17 -
EUDNet
• Research contrasts between (EU)EUD and
(US)EUP/EUSE
• European projects are more likely to:
– target users in specific technical domains
– study organisational issues
– address adults, not children / students.
End-User Software Engineering SIG at CHI’2008
- 18 -
Breakout Groups
• Groups:
– Design & Requirements (Alan Blackwell)
– "Create" (Brad Myers)
– Reuse (Mary Beth Rosson)
– Testing & Verification (Margaret Burnett)
– Debugging (Andy Ko)
• Focus on the S/W Eng issue and how it applies to each of the
different domains
– Domain-driven compare/contrast
– What has been done relevant to that domain for this S/W Eng issue
– Which domains are most challenging and easiest?
– How similar and different
• Take notes in PowerPoint for 3 minute summary
End-User Software Engineering SIG at CHI’2008
- 19 -
Industry Attendees at CHI’2008 SIG
•
Paula Dimaio – from media, consultant: integrated methodology
S/W+UE for practitioners. Ontology engineering
•
Marcus Stolze – IBM TJ Watson research lab. Business analysts,
enterprise applications, dynamic access, business
•
Jack Zaintz – SOAR Technology, modeling and simulation for military,
let end users do their own modeling, training operators – adaptive
scripting of models
•
John Richards – IBM TJ Watson – support business analysts and
architects when doing sense-making. Also, petaflop with 13-fold
increase in programmer-productivity
•
Claude Knaus – indep. developer: interaction design to create s/w
dev. Tools
•
Laura Beckwith – Microsoft Visual Studio team: middle between
professional and EUP, e.g., consultants who want to make money off
of programming
•
Steve Clarke – also with MS VS
•
Tessa Lau & Allen Cypher – IBM Almaden – Co-Scripter: EUP for web
•
Steve Pemberton – Designer of XForms
End-User Software Engineering SIG at CHI’2008
- 20 -
Domains
•
Web Applications
•
Education, pen-based application: primary school (6-10years old)
•
Coaches of sports teams
•
Enterprise users (all employees)
•
Ambient user interfaces – in home, consumer electronics
•
Visualization for air traffic controllers (ATC), operators monitoring
computer systems, UAV path has to be designed – has a language
w/conditionals but want a UI
•
Modeling & simulation for military exercises for military operations,
NASA telemetry engineers
•
Professional programmers who act like EUP
•
Analysts who design and build enterprise applications, how can they
provide more of the implementation – let them design their own UIs.
As part of WebSphere & Rational
•
Personal Information Management
End-User Software Engineering SIG at CHI’2008
- 21 -
“Create” group results
•
Jack: Lots of different tools that have to be used together
– Large Simulation System (JSAP) – config. files, UIs for scripting movement,
intelligent entity control IDE (SoarTAC) but not for EU.
– Simplified state machines – IDEs but not very good, FSM editor but
pretends to not be a graph
– Better tools for Intelligent Entity Control – textual IDEs w/some graphics for
FSM, graphic forms
•
Flash, Frontpage,
– Make the web sites conform to accessibility guidelines
•
Educators
– How to learn Chinese, Geometry, Mathematics
– Low-level platform “UI Widgets”: Text, Ink, Math formulas, Geometry,
Chemistry
– Pen-based design tool for designing functions they like
– Interactive, drag-and-drop tool, set properties
•
Sport fields
– Different coaches have similar needs, but different formats
– Different kinds of spreadsheets
End-User Software Engineering SIG at CHI’2008
- 22 -
“Create” group results
• Olivier Liechti – simple for interactive remote control
for consumer electronics using Visual Programming,
on handheld display
• CoScripter – IDE for scripting on the web, mostly by
PBE
• Enterprise Applications, start in an unorganized
fashion – how turn spreadsheet into a real
enterprise applications?
– Describe their process with an XML language
– Structured applications around the data objects
– State machines, specifying operations to change states,
describe choices, process and focusing on data and states
of the data. Data flow.
– State of data objects, focus on access control questions
– Additional aspect: input validation – separate framework
needed
End-User Software Engineering SIG at CHI’2008
- 23 -
“Create” group results
• Max: PIM: scripting language (in JavaScript)
– Specifying conditions under which things are
triggered by example – go into past and select a
moment when would want something to happen
– Select things that are relevant in the real user
interface of Web-based PIM
– Triggered on what documents are open, what
music listening to, Twitter status (location)
– Actions: call any web-based service, desktop
services (turn on music, etc.)
• Entry-level JavaScript programmer
End-User Software Engineering SIG at CHI’2008
Categories of Creating
• Drag-and-drop
• Visual programming
• PBE
– Textual model that can read but not edit
– Hidden Markov Models
– Creates conventional script
• Textual
– Copy-and-paste: different style of programming
– IDE:
• Code completion, automatic error finding and fix suggestions
makes them much better for EUP
– Exploration tool: PIG tool from Berkeley simulates as you go
• Form-filling – mail filters, Jack’s IFOR (intelligent forces)
End-User Software Engineering SIG at CHI’2008
- 25 -
Testing and Verification Group
•
Haven’t seen different work styles.
•
Steven studied concept of a tester. How can I break? Study of people whose job responsibitilities were testing.
–
•
Some were “testers”. Some were “business analysts”, “programmers”. All did the same, namely black-box.
Piero: 1. Medical, 2. Mech engr, 3. Software people. They have different views on testing, have different procedures for
testing.
–
Eg, people from physics and medicine -- they have experience testing physical machines but not software. They do very “black box”
approaches. (Often, their software is part of the machines they buy.)
•
Steven has seen a lot of black-box.
•
Sometimes end users don’t think about testing at all, at least in spreadsheets.
–
•
Maybe because of the immediate visual feedback -- they may in some sensre regard it as “already tested”.
Oracle problem -- making errors in judgments of failed/successful tests.
–
Overconfidence is a problem with end-user programmers.
–
With images it can be harder to evaluate the answer.
•
Under the hood: software engr. talks about the underlying logic.
•
Code inspection.
–
Gender differences (see 11:30 session).
•
Assertions.
•
Scenario-based approach: “When a user does this, then that, the following happens”. A storyboard.
•
–
A semi-quote: “Developers take the happy path. I take all the other paths.”
–
We wonder: is this true mostly in imperative paradigms, or would it also apply to more declarative ones?
Interactive systems have important testing issues in their own right,
–
because end-user “HCI” work to customize these could lead to terrible errors.
End-User Software Engineering SIG at CHI’2008
- 26 -
Testing and Verification Group Results
• Haven’t seen different work styles.
• Steven studied concept of a tester. How can I break? Study of
people whose job responsibitilities were testing.
– Some were “testers”. Some were “business analysts”,
“programmers”. All did the same, namely black-box.
• Piero: 1. Medical, 2. Mech engr, 3. Software people. They have
different views on testing, have different procedures for
testing.
– Eg, people from physics and medicine -- they have experience
testing physical machines but not software. They do very “black
box” approaches. (Often, their software is part of the machines
they buy.)
• Steven has seen a lot of black-box.
• Sometimes end users don’t think about testing at all, at least in
spreadsheets.
– Maybe because of the immediate visual feedback -- they may in
some sense regard it as “already tested”.
End-User Software Engineering SIG at CHI’2008
Testing and Verification Group Results
•
Oracle problem -- making errors in judgments of failed/successful
tests.
– Overconfidence is a problem with end-user programmers.
– With images it can be harder to evaluate the answer.
•
Under the hood: software engr. talks about the underlying logic.
•
Code inspection.
– Gender differences (see 11:30 session).
•
Assertions.
•
Scenario-based approach: “When a user does this, then that, the
following happens”. A storyboard.
– A semi-quote: “Developers take the happy path. I take all the other paths.”
– We wonder: is this true mostly in imperative paradigms, or would it also
apply to more declarative ones?
•
Interactive systems have important testing issues in their own right,
– because end-user “HCI” work to customize these could lead to terrible
errors.
End-User Software Engineering SIG at CHI’2008
Design & Requirements Group
Results
• Who “interprets” the requirement?
• Normally a professional translates from messy/concrete to
the computer abstraction.
• Supply side: But without professional? Can a tool provide
the trust
• Demand side: Two kinds of attitude to abstraction:
– Some expert communities have good domain abstractions
– Some communities share concrete examples, don’t explicitly
discuss abstraction
• Alternative: Focus on community (content, analogy,
sharing)
End-User Software Engineering SIG at CHI’2008
Reuse Group Results
• Gerhard Fischer (U Colorado), Dane Bertram (U Calgary),
Allen Cypher (IBM Research), Craig Anslow (Victoria Univ.
Wellington), Laura Beckwith (Microsoft), Tom LaTosa
(CMU), Chris Scaffidi (CMU), Volkmar Pipek (U Ziegen)
• Domain-based comparisons
– Content domains versus paradigm domains
• E.g. web, scripting, DB versus business, science, etc.
– May also be something about where in lifecycle
• Find & adapt
• Create and organize
• Inherit and learn
• Collaborative/mutual discovery
End-User Software Engineering SIG at CHI’2008
Domain contrasts
•
Business processes (scripting)
–
•
•
Information providers, e.g. admin who has content to post
–
Example might be using Lotus Notes / Dominoes;
–
Observed lots of example/edit based reuse
–
This could include engineers
Consultants, integration including some programming
–
•
Just want a solution demonstrated ASAP, find and connect code together
Java professional programmers
–
•
Oriented toward large enterprises that have shared practices and thus built-in
motivation for sharing
Reuse within maintenance paradigm, finding model in others’ code
Meta level, building tools that are situated within a domain
–
Human problem domain interaction --> domain-oriented design environments
–
Professional programmer builds the blocks to enable domain specialist
–
Assumes goal of “reflective communities” that include collaborations
End-User Software Engineering SIG at CHI’2008
Debugging Group
• Debugging across domains
• Examples of end-user programming debugging:
– large scale defense systems
– multi-sensor data fusion
– mashups of web resources
– programming by example
– natural language programming
– debugging intelligent user interfaces
End-User Software Engineering SIG at CHI’2008
- 32 -
Debugging Group
• One problem with debugging mashups is that the tools
haven't caught up with the state of the art commercial tools
for more popular debugging platforms.
• Debugging multi-sensor systems is difficult because they
often involve complex AI and machine learning algorithms.
• Perhaps the debate between static and dynamic testing
would be moot if we had better tools for understanding the
dynamic behavior of program behavior.
• Almost always trust the sensors, but almost never trust data
sources in web mashups.
• Henry's waterlanguage.com is trying to create a unified
platform for web development.
End-User Software Engineering SIG at CHI’2008
- 33 -