[SCC_Active_Members] Source Code Searching

[Courtney, Lee]

Henry et al,

> " As far as we know, no-one in the world has created 
> subroutines to extract useful search indices from program 
> collections.  Doing this well might be a nontrivial 
> challenge.  I hope to find volunteers to help tackle it, 
> starting with obvious heuristics.  (If we can invent useful 
> algorithms, programming them as GSDL plugins will be neither 
> difficult nor laborious.)"

Much work has been done in this area for many many years. The Mining
Software Repository Workshop series
(http://msr.uwaterloo.ca/msr2007/index.html) has some information that
is relevent to this discussion. I'd also look at eclipse.org as an area
where some interesting tools and analysis paradigms are being developed.


One point we need to emphasize is the work of SPG is very relevent to
CURRENT efforts in the area of mining software repositories and
attempting to enhance the productivity of software development teams
thru the use of reuseable modules and componentized development
methodology. Leveraging an existing tool into a GSDL plug-in to analyze
software is a great idea. However, from attending MSR Workshops and
following (a little of) the literature a lot of the work is around
contemporary systems (e.g. CVS, C programming language). I suspect a lot
of the algorithms and approaches can be borrowed, but actual
implementation will need to be expanded to accommodate a more robust
landscape of problems.

Cheers,

Lee Courtney
Product Line Manager - Linux for Consumer Devices
Wind River
500 Wind River Way
Alameda, California 94501
 
  Office: 510-510-2763
    Cell: 650-704-3934
Yahoo IM: charlesleecourtney
 

> -----Original Message-----
> From: scc_active-bounces at mail.computerhistory.org 
> [mailto:scc_active-bounces at mail.computerhistory.org] On 
> Behalf Of Randall Neff
> Sent: Monday, June 25, 2007 10:27 AM
> To: SCC_active at mail.computerhistory.org
> Subject: [SCC_Active_Members] Source Code Searching
> 
> from: H.M. Gladne
> " As far as we know, no-one in the world has created 
> subroutines to extract useful search indices from program 
> collections.  Doing this well might be a nontrivial 
> challenge.  I hope to find volunteers to help tackle it, 
> starting with obvious heuristics.  (If we can invent useful 
> algorithms, programming them as GSDL plugins will be neither 
> difficult nor laborious.)"
> 
> 
> 
> There are a number of startups and Google that have source 
> code searching capabilities, usually on open source software:
> 
> 
> www.koders.com
> www.krugle.com
> www.codase.com
> www.allthecode.com
> www.google.com/codesearch
> 
> The problem is that they are based on just strings in the 
> source (including comments), and nothing on higher level 
> structure.  So if I look for 'graphic editor", the search 
> returns classes and java packages named "graphicEidtor" 
> "grahic_Editor", etc, and not a collection of classes or a 
> complete program that would comprise an editor.
> 
> 
> Randall Neff,
> _______________________________________________
> SCC_active mailing list
> SCC_active at mail.computerhistory.org
> http://mail.computerhistory.org/mailman/listinfo/scc_active
>