16: Team Project 16: Intelligent Diff Meeting Date: 09/19/2005 Participants: Client: Lead: Ops Concept: Proto-typist: Architect: Modeler: Contents: 1. 2.
Professor Brown Robert Huber Sunil Raga Dave Naumann Tom Ackenhausen Kevin Sigmund
Meeting Minutes Summary Tasks List
Meeting Minutes: 1. Boeing Laptop Reservation Website: a. http://wpscsc.web.boeing.com Current CodeCount Deficiencies: a. CodeCount counts all code, not just code that a person actually writes; CodeCount falls short on reuse code counting. b. CodeCount falls short on accounting for moved (re-factored) code; CodeCount counts all code in files but does not identify between code that was moved from one place to another. Current CodeCount Needs: a. CodeCount needs to account for more than just new vs. deleted code. b. CodeCount needs to account for re-factored code. Driving Needs (COCOMO II): a. COCOMO II has some idea about Adaptation type code (reuse code). For each module in COCOMO II double-click on Module Size field then select the Adaptation radio button. The dialog box that comes up has the following fields: i. Initial SLOC Count ii. % Design Modified (DM) iii. % Code Modified (CM) iv. &Integration Modified (IM) v. Software Understanding (SU) vi. Assessment & Assimilation (AA) vii. Unfamiliarity with Software viii. % Components Automatically Translated (AA) ix. Automatic Translation Productivity (ATPROD) x. Breakage (BRAK) The general idea is to improve these entries by providing SLOC information at the project (not file source-level) to the user (formatted output file) on SLOC deleted, new, modified moved, modified unmoved, and unchanged. Project Idea: a. The core solution is to take two CM versions of a project (full set of source files) and build an intelligent diff that separates the code into files that represent deleted, new, modified moved, modified unmoved, and unchanged code categories and then feed these files to CodeCount. Lastly take the CodeCount output and then reformat this to a userfriendly level that (if possible) correlates to the Adaptation dialogue of COCOMO II. Project Scoped Levels for Diff Intelligence a. Level 1 Intelligent Diff i. Given an original and file, its final version separate code as follows, note that these track only changes within a single file by using two versions of that same file: 1. Deleted (Not re-factored, pure deletion) 2. New (Not re-factored from elsewhere, pure new) 3. Modified (Not Moved, and changed) 4. Modified Move (Moved, and changed) 5. Unchanged (Not moved, and not changed) 6. Unchanged Move (Moved, and not changed) b. Level 2 Intelligent Diff i. Given a user provided re-factoring map, a table that lists source code blocks (by start/end line numbers) with source file-names and the corresponding destination
2.
3.
4.
5.
6.
code blocks with destination file-names, count the re-factored code as follows, note that these track only changes between 2 files using the re-factoring map: 1. Unchanged Move Between 2 Files 2. Modified Move Between 2 Files c. Level 3 Intelligent Diff i. Automate the discovery portion of Level 2 Intelligent Diff, that is replace the need of a user provided re-factoring map with discovery software (essentially automate the generation of the re-factoring map).
7.
Prototypes: a. Level 1 Intelligent Diff
Original File
Final File
Deleted Unchanged Modified
New Unchanged Modified
Level 1 Intelligent Diff
Deleted (Not re-factored, pure deletion) File New (Not re-factored...