ExampleWalkthrough

We demonstrate extracting try/catch blocks from a small Java project into a postgresSQL database, stored locally. We assume an installation of PostgreSQL, Java, Python, and the required li- braries. For this demonstration we will interact with the Post- greSQL database via pg3admin. Below we provide a textual walk- through of this task, but we will include a video demo shortly that will be linked in the README at www.github.com/caseycas/ gitcproc.

1. Assume we have a database called caseycas. Create a new schema in the database, called java_error_handlers to store our logs.
	• In pg3admin, we right click on the Schemas tab in the Object browser, and select New Schema...
	• We fill in the name section with java_error_handlers, and the owner section with our username, in this case caseycas, then hit okay. We are now done with the database setup.

2. Create a new config.ini file, called demo_conf.ini. We will store this in gitcproc/src/util for the moment.

3. Open the config.ini file and fill in the sections. "Walkthrough.ini" shows an example configuration for this task. [Database] hosts our database connection information, with our database name, username, host, listening port, schema and output table names. [Repos] contains the file name of our list of projects to down- load and the storage location where they will be downloaded. [Keywords] links to our keyword file. [Log] is where we specify what languages to parse. [Flags] includes several flags for our output types and debug level. There we have DATABASE set to 1 for output, and the DEBUG and DE- BUGLITE options set to 0. A more detailed explanation of each variable in the file can be found in the README file in our repository: www.github.com/caseycas/gitcproc.

4. Save the changes to .ini file.

5. Create the file demoKeywords.txt in gitcproc/src/logChunk. This is the keywords file linked in the [Keywords] section of the .ini file. Note that all paths in the .ini are relative to where you execute the main script from, which is currently gitcproc/src/logChunk.
	• Fill in the first line with "try", included, block
	• Fill in the second line with "catch", included, block

6. Save the keywords file.

7. Create the repository file. For this demo’s time purposes, we will study just one project, the language Clojure, which is implemented partially in Java. Add the line clojure/clojure.

8. Save the repository file.

9. Now, we can run the tool. The master script, gitcproc.py is located in gitcproc/src/logChunk. The script has several options, detailed in the README or by supplying the script with the -h option. We will run the tool with the command python gitcproc.py -d -wl -pl ../util/demo_conf.ini. This will download the project, write out the git log, parse it and put it into the database. After a few seconds, the tool will prompt you to input your database password.

10. Once the tool finishes running (it should take no longer than a few minutes), we can check the results in the tables the tool created. The table name provided under the .ini option table_change_summary: contains commit level information; the table name listed under table_method_detail: contains the changes to functions and aggregated blocks of non change functions.

	• Try running the sql command "SELECT * FROM java_error_handlers.method_change_detail WHERE try_adds > 0 or try_dels > 0". This will show all func- tion changes where code inside a Java try block was modified.
	• The data can now be used to discover the evolution of try/catch code in the Clojure project.