Bonsai Issue, October 3rd
In order to progress with the project as outlined in here, there are a few matters to be resolved. In order to properly retrieve Mozilla source files from various timeframes and versions, team needs to have sufficient understanding on how bonsai works. It is assumed that once the query process is performed and a sample of the output is retrieved, the team can proceed with the design of the prototype.
Here are the current concerns to date:
- Bonsai checkin -- The team needs to understand how cvsqueryform.cgi and cvsquery.cgi in the bonsai page handles client inputs, especially regarding the timeframe (e.g. in the last # hours, day, week, etc.). Also, how does the output from bonsai really look like? Is it a text file? Can this output be used to directly download the source files from CVS?
- Bonsai version history -- When user asks for a certain file from the source tree, bonsai returns all versions of that file. How does bonsai handle such request, and how does the output look like?
- Bonsai differences -- When user clicks on a certain version of a given file, bonsai displays that file and displays all changes within the file from previous version, properly formatted and highlighted. Again, how does bonsai handle such request, and how does the output (originally) look like?
The team's idea is to come up with a tool that forwards a query to bonsai independent from the bonsai query site, retrieve the outputs, and work with them. Once the team manages to download the previous (working) version, delta debugging can be applied to the crashed version to remove one changeset after the other, until the real cause of the problem is located.