blob: 5a2872f9ab275a3adc4ae0e5d438d8115fc2f1fa [file] [log] [blame]
================= NOTES TO OURSELVES ================
_______ RUNNING _______
To run the batch or incremental:
cd ~bfreeman/org.eclipse.dash/commits-explorer
sh update.sh
nohup sh run-incremental.sh &
To force a batch run:
nohup sh run-batch.sh &
Log files:
The insert log file is ~bfreeman/org.eclipse.dash/commits-explorer/logs/
_______ SWITCHING TO CRON _______
We need to switch to using cron to run this. To do so, we need to
solve the ssh problem (not needing a password).
_______ INSTALLING _______
1. Apache, PHP, Perl, MySQL; set query caching to 32Mbs or more
2. create the DBPASSWORD.txt file in the batch sub-directory; edit
the dbpassword.php file in the web-api directory to reference the
batch/DBPASSWORD.txt file
3. cron to run batch/run-incremental.sh once per day
4. web-api and web-app directories exposed through Apache
There are a few web-apis on www.eclipse.org/projects/web-api/ to support
these programs.
_______ FUTURE WORK _______
Haven't resolved what goes wrong with the last of the packages
of tools.cdt in the rlog parsing. Perhaps something with including
dashed lines in comments?
We also had a problem parsing org.eclipse.ui.navigator - same problem?
Use wikitect to generate the project mappings and thus allow/distribute
the maintenance to the architecture council.
Perhaps use a wiki page to maintain the type mappings.
Another wiki page idea: have a way to people to fix the bad data
issues about themselves. For example, a way to annotate that their
login id changed from darinw to dwright (i.e., that both are the same
person), and a way to annotate that person X worked for company Y
during one period and then company Z for another period, etc.
Data cubes (OLAP) often examine more than just number of transactions.
Thus one thing we could add is instead of COUNT(*) in our summaries to
do SUM(CHANGE_SIZE).
If suffixes (file types) are longer than N (12?) translate them to "none".
Tag cloud generation from comments. Animated tag clouds from comments
over time.
Include website commits from /cvsroot/org.eclipse
Do both # of commits and # of lines of change.
Keep track of data valid time and show it on the web page "data collected on XX"