Developers come and go but the code remains About me Committer for - - PowerPoint PPT Presentation

developers come and go but the code remains about me
SMART_READER_LITE
LIVE PREVIEW

Developers come and go but the code remains About me Committer for - - PowerPoint PPT Presentation

Developers come and go but the code remains About me Committer for PhD from + CTO of About us .com supported by What went on in my ex-colleagues head? David Monniaux How can I help the the new guy become productive?


slide-1
SLIDE 1

Developers come and go but the code remains

slide-2
SLIDE 2

About me

CTO of PhD from + Committer for

slide-3
SLIDE 3

.com

supported by

About us

slide-4
SLIDE 4

“What went on in my ex-colleague’s head?”

David Monniaux

slide-5
SLIDE 5

“How can I help the the new guy become productive?”

slide-6
SLIDE 6

“How can I help the the new guy become productive?”

slide-7
SLIDE 7

It takes a new developer three to six months to become fully productive.

[De Marco, Peopleware: Productive Projects and Teams, ISBN:0932633439]

slide-8
SLIDE 8
slide-9
SLIDE 9

“Where to get started?”

slide-10
SLIDE 10
slide-11
SLIDE 11

On average, developers spend 25 seconds in the code completion dialog.

[Scheller & Kühn, TU Wien, IASTED ’13, DOI:10.2316/P .2013.796-027]

slide-12
SLIDE 12
slide-13
SLIDE 13

Like every good dinosaur...

David Monniaux

slide-14
SLIDE 14

Jerry Daykin

...every developer leaves a trail in the code.

slide-15
SLIDE 15

Jerry Daykin

And if you take a step back, patterns emerge.

slide-16
SLIDE 16
slide-17
SLIDE 17
slide-18
SLIDE 18
slide-19
SLIDE 19

Our own dog food

Nancy Gonzalez

slide-20
SLIDE 20

Our own dog food

slide-21
SLIDE 21
slide-22
SLIDE 22

jfxrt-2.2.1.jar jfxrt-8.0.0.jar jfxrt-2.x-call.zip jfxrt-8.x-call.zip

Samples APIs Models Proposals

slide-23
SLIDE 23
slide-24
SLIDE 24
slide-25
SLIDE 25

samples++

slide-26
SLIDE 26

samples++

slide-27
SLIDE 27
slide-28
SLIDE 28
slide-29
SLIDE 29

Demo time

slide-30
SLIDE 30

Get the Developer Preview

www. .com/download

slide-31
SLIDE 31

Free as in

Alexandre Lazaro

slide-32
SLIDE 32

See Mine for yourself

slide-33
SLIDE 33

Bring your own code

@ Eclipse Code Recommenders table

slide-34
SLIDE 34

samples = 1

slide-35
SLIDE 35

samples = 0

slide-36
SLIDE 36

“No samples?”

slide-37
SLIDE 37

“No samples?”

slide-38
SLIDE 38

“Learn...”

slide-39
SLIDE 39

“...as you type”

slide-40
SLIDE 40

Feedback!

slide-41
SLIDE 41

Feedback!

slide-42
SLIDE 42

Feedback!

slide-43
SLIDE 43

Feedback!

slide-44
SLIDE 44

It’s science!

1/4 1/2 3/4 1 SWT MRR

Eclipse Frequency-based ML-based

slide-45
SLIDE 45

It’s science!

1/4 1/2 3/4 1 Java SE MRR

Eclipse Frequency-based ML-based

slide-46
SLIDE 46

Code completion based on implicit user feedback

Sven Amann, FB 20, Computer Science (Master of Science) Technische Universität Darmstadt Department of Computer Science Software Technology Group Reviewer: Prof. Dr.-Ing. Mira Mezini Advisor: Dr.-Ing. Marcel Bruch and Dr.-Ing. Andreas Sewe 31.03.2013
slide-47
SLIDE 47

http://build.codetrails.com/updates/theses/amann/

slide-48
SLIDE 48

On the horizon

Adam Wyles

  • Crowd-sourced completion in-the-wild
  • Intelligent type completion
  • More detailed models
  • Reduced memory consumption
slide-49
SLIDE 49

From A to ...

slide-50
SLIDE 50
slide-51
SLIDE 51

} not yet used

slide-52
SLIDE 52

More detailed models

AST

Expected type

Superclass

Annotations

I n t e r f a c e s

Overridden method

C u r r e n t p r e fi x

Developer

Recent/open files

Names

Natalie Lucier

slide-53
SLIDE 53

Nick Perla

Reduced memory consumption

slide-54
SLIDE 54

http://eclipse.org/recommenders/

@recommenders

slide-55
SLIDE 55

Q & A

slide-56
SLIDE 56

Give Feedback on the Sessions

1

Sign In: www.eclipsecon.org

2

Select Session Evaluate

3

Vote

Code Recommenders: Developers come and go but the code remains

Andreas Sewe