Challenges in Code Discoverability: or: life after - - PowerPoint PPT Presentation

challenges in code discoverability
SMART_READER_LITE
LIVE PREVIEW

Challenges in Code Discoverability: or: life after - - PowerPoint PPT Presentation

Challenges in Code Discoverability: or: life after google.com/codesearch? Peter Teuben (UMD) Peter Teuben (UMD) Allice Allen (Calverton, MD) Allice Allen (Calverton, MD) Antikythera Robert Nemiroff (MTU) Robert Nemiroff (MTU) Lior Shamir


slide-1
SLIDE 1

Challenges in Code Discoverability:

  • r:

life after google.com/codesearch?

Peter Teuben (UMD) Allice Allen (Calverton, MD) Robert Nemiroff (MTU) Lior Shamir (LTU) Peter Teuben (UMD) Allice Allen (Calverton, MD) Robert Nemiroff (MTU) Lior Shamir (LTU)

Long Term Preservation of Analysis Capabilities in Astronomy Software Antikythera

slide-2
SLIDE 2

Papers

 Weiner et al.(2009) whitepaper for 2010 decadal

 Software wants to be free ; Manifesto

 Shortridge (2009) – 2008 ADASS – AstroShare

 Talk amongst yourselves

 Gray & Mann (2011) – 2010 ADASS BoF

 Need our own journal?

 Grosbol & Tody (2010) -

 More efficient access to software

slide-3
SLIDE 3

Google Code Search (R.I.P.)

slide-4
SLIDE 4

Scientific Codes

 Code Types: (very heterogeneous)

 Telescopes Control  Data Reduction & Pipelines  Data Analysis  Data Modeling

 Code Correctness

 Regression Testing  Reproducibility, Verifiability

 Code Sharing

slide-5
SLIDE 5

Brief History of Codes

 WWII bomb codes, opacity tables  Wilson-Devinney (eclipsing binaries)  Aarseth (N-body)  DAOPHOT, Sextractor  GALFIT  GADGET 1,2,3, Arepo

slide-6
SLIDE 6

Then came the internet

slide-7
SLIDE 7

Community Examples

 freshmeat.net (now: freecode.com)

 e.g. HDF

 sourcecode.org

 e.g. Sextractor, Stellarium

 github.net (w/ SCM)

 e.g. s2plot

 google.code.com (w/ SCM)

 e.g. CASA core

slide-8
SLIDE 8

Efforts in Astronomy

 AstroWEB (1994 Jackson/Wells/Adorf)

 Different views of the same database, hosted at 7

different locations (NRAO still alive!)

 ASDS (-1998; Hanisch et al.)  AstroForge (2003-2005; Brunner et al.)

 Modeled after sourceforge

 SkySoft (2001-; Baffa, Giani & Checcucci)  ASCL v1 (1999-2002 ; Wallin & Nemiroff)

slide-9
SLIDE 9

Astronomy Efforts (cont'd)

 IAU endorsed code repositories

 SOFA (coordinates) + validation suite

 AstroCode (2006-2011; Moore)  Astro-Sim  AstroShare (2008- ; Shortridge)  Many many meta listings (e.g. wikiversity, ...)

slide-10
SLIDE 10

ASCL v1

 March 1999 – Feb 2002  37 codes from/used in peer reviewed papers  How many survived:

 17 have still working URLs (1 redirect)  17 have dead URLs  1 has dead URL, but code download works  2 were only on ASCL v1  33 are now in ASCL v2

slide-11
SLIDE 11

ASCL v2

 published codes

 340 342 345 published codes, existing publications

 Uses phpBB (off APOD)

 Discussion threads  Each Discussion is a Code Entry

 Advisory Committee

 Teuben (chair), Nemiroff, Wallin, Mink, Berriman,

Shamir, Hanisch,... and

 Allen (editor)

 White Paper – in prep.

slide-12
SLIDE 12
slide-13
SLIDE 13

ASCL.NET

slide-14
SLIDE 14
slide-15
SLIDE 15

Which Codes to ”Archive”

By ”Support” Levels?

 0 = with regression / validation suite  1 = peer reviewed journal  2 = conference proceedings style  3 = web site  4 = word of mouth

slide-16
SLIDE 16

Other

 Literate programming

 CWEB (Knuth)  Amrita (www.amrita-ebook.org)  CDF (Wolfram Research)

 Algorithm detectors  Cross-indexing

slide-17
SLIDE 17

Future

 Widen the scope?

 Part of the IVOA?  ADASS?

 cf. BoF

 we reference the ”methods” and the ”data” - Derriere

 Repository? - cf. AstroForge

 freshmeat code.google.com github (”AstroForge”)  Google summer of code: astrometry.net

 ASCL v3???

slide-18
SLIDE 18
slide-19
SLIDE 19

Papers

 Weiner et al.(2009) whitepaper for 2010 decadal

 Software wants to be free Manifesto

 Shortridge (2009) – 2008 ADASS – AstroShare

 Talk amongst yourselves

 Gray & Mann (2011) – 2010 ADASS BoF

 Need our own journal?

 Tody & Grosbol (2010) -

 More efficient access to Software

slide-20
SLIDE 20

Audience Poll Show of hands:

1) Have you had your code cited in a paper? 2) who has code that fits in the current ASCL model?

slide-21
SLIDE 21

Conclusions

     

slide-22
SLIDE 22

Why (not) release source code?

 Quick and Dirty  Intellectual property

 Employee Lawyers – don't ask don't tell

 Not career advancing

 20/80 rules of development/maintenance

 Produce wrong results  Job Security?  Reimplemention always fits better

slide-23
SLIDE 23

Why (not) release source code?

 Observatory Code Syndrome

 ”specific to my observatory”

slide-24
SLIDE 24

Software Museum?

 Curation?  Orphaned code  Examples: NEMO

 Aarseth codes, making Plummer sphere

 Code evolution:

 AIPS, SDE, MIRIAD, AIPS++ CASA  Gadget1, 2, 3, Arepo

slide-25
SLIDE 25

Previous Efforts (practical)

 LfA - Linux for Astronomy – software on CD  Skysoft - ESO's Mac compilation (Hook et al.)  Packaging:

 SDSS group – perl based (Lupton)  Astromake – shell based (Teuben)