Forgot your password?
SPIP Work Plan
Document Actions
Tasks:
Analyses
1. Bug Lifetime analysis.
-------------------------
- SF Analysis done and reported in ICSE workshop paper (Towards ...)
- Haven't run this analysis for Savannah and Apache Bugzilla.
- need to run appropriate database queries against the databases on floss.syr.edu
- then run same R-script analysis that Kevin did for SF.
2. Code re-use.
----------------
- Need to get dependancy structure of Debian and Gentoo. Deborphan a possibility here. Could also write to NathanLaBelle and Eugene Wallingford who did the debian structure paper (not as a success measure)
- Work out what the appropriate graph theoretic (SNA) measurement is. Possibly prestige? Or a simple count of the number of times a package is re-used (at any level of the tree).
- the simple count can be done using
: shangorilla:~# apt-cache rdepends libc6 | wc -l
8794
Getting the reverse dependancies for all packages seems to be done by this:
apt-cache dumpavail | grep ^Package: | cut -d: -f2 | xargs apt-cache rdepends > package-reverse-deps.txt
but rdepends lists conflicts and suggestions. Anyway it can be worked out from this data. Awesome.
3. Developer and active user attraction and retention.
-----------------------------------------------------
- examine data snapshot data from OSSMole (which is not all in the database I'm afraid)
- Count the number of leaves/joins (of developers) in each time period using the data from the summary pages. in SF and savannah.
- possibly a more complex count of new active users (ie posted to the bug tracker) Would want to gauge tenure somehow, did they stick around. Could make a sequence diagram ..
- probably better to simply count the number of active users in each time period. Is it going up or down? That can be done for 140 projects from Sourceforge and all the projects in Apache Bugzilla and Savannah.
Reading + Literature search
---------
Need to review Katherine Stewart's work (at least)
Found a few other good mainstream IS success papers.
Review floss papers (on opensource.mit.edu) and code their dependant variables or use of success
Writing
--------
Need to expand the abstract out like:
- highlight existing measures
- show their fit, or lack of, with FLOSS
- critique the current measurement of Success in FLOSS.
- explain theorectical model (from HICSS) which supports the team paper
- show our adapted general model of success (something in Kevin's slides, that is not in the papers ...), point out where existing measures can fit in.
- use the interviews and slashdot focus group for validation of this model.
- so we need to measure things about ... which is why we need new measures.
- the three measures
- way each is relevant
- method and results for each
- interactions between our measures
- conclusions.
Drop in appropriate bits from ICIS, ICSE, HICSS papers
Analyses
1. Bug Lifetime analysis.
-------------------------
- SF Analysis done and reported in ICSE workshop paper (Towards ...)
- Haven't run this analysis for Savannah and Apache Bugzilla.
- need to run appropriate database queries against the databases on floss.syr.edu
- then run same R-script analysis that Kevin did for SF.
2. Code re-use.
----------------
- Need to get dependancy structure of Debian and Gentoo. Deborphan a possibility here. Could also write to NathanLaBelle and Eugene Wallingford who did the debian structure paper (not as a success measure)
- Work out what the appropriate graph theoretic (SNA) measurement is. Possibly prestige? Or a simple count of the number of times a package is re-used (at any level of the tree).
- the simple count can be done using
: shangorilla:~# apt-cache rdepends libc6 | wc -l
8794
Getting the reverse dependancies for all packages seems to be done by this:
apt-cache dumpavail | grep ^Package: | cut -d: -f2 | xargs apt-cache rdepends > package-reverse-deps.txt
but rdepends lists conflicts and suggestions. Anyway it can be worked out from this data. Awesome.
3. Developer and active user attraction and retention.
-----------------------------------------------------
- examine data snapshot data from OSSMole (which is not all in the database I'm afraid)
- Count the number of leaves/joins (of developers) in each time period using the data from the summary pages. in SF and savannah.
- possibly a more complex count of new active users (ie posted to the bug tracker) Would want to gauge tenure somehow, did they stick around. Could make a sequence diagram ..
- probably better to simply count the number of active users in each time period. Is it going up or down? That can be done for 140 projects from Sourceforge and all the projects in Apache Bugzilla and Savannah.
Reading + Literature search
---------
Need to review Katherine Stewart's work (at least)
Found a few other good mainstream IS success papers.
Review floss papers (on opensource.mit.edu) and code their dependant variables or use of success
Writing
--------
Need to expand the abstract out like:
- highlight existing measures
- show their fit, or lack of, with FLOSS
- critique the current measurement of Success in FLOSS.
- explain theorectical model (from HICSS) which supports the team paper
- show our adapted general model of success (something in Kevin's slides, that is not in the papers ...), point out where existing measures can fit in.
- use the interviews and slashdot focus group for validation of this model.
- so we need to measure things about ... which is why we need new measures.
- the three measures
- way each is relevant
- method and results for each
- interactions between our measures
- conclusions.
Drop in appropriate bits from ICIS, ICSE, HICSS papers
