COMP9315 24T1
Assignment 1
DBMS Implementation
Testing Resources
Last updated: Friday 23rd February 1:51pm
Most recent changes are shown in red ... older changes are shown in brown.
This document describes a testing framework for determining whether your implementation of the PersonName data type is correct. This testing framework contains a subset of the tests that will be made in auto marking. If you pass all the tests here, you ought to pass the majority of the auto-marking tests. Note, however, that passing all of these tests does not guarantee that you will pass all of the auto-marking tests.
The first step in setting up the testing framework is to make a testing directory under your vxdb local storage:
|
$ ssh nw-syd-vxdb $ source /localstorage/$USER/env $ mkdir /localstorage/$USER/testing |
The next step is to set up the testing infrastructure:
|
$ cd /localstorage/$USER/ $ tar -xf /web/cs9315/24T1/assignments/ass1/testing/testing .t |
This sets up a number of files and directories under the testing directory. Note that some of these are actually symbolic links to our files so that (a) you can't change them, and (b) you save storage space under your own directory.
The tar command creates (among other things) the following:
run_test.py a Python script that will run all of the tests; use this after you have gotten your implementation of PersonName into a reasonable state; this is a symlink to a script in the COMP9315 account
a tests/ subdirectory, containing all of the test files; this is a symlink to files under the COMP9315 account, so all of the tests are read-only unless you copy them manually to your testing/ directory
There are three schemas provided with the testing framework, and each schema has corresponding data sets.
|
Users(realname::PersonName) |
|
Students(id::int, name::PersonName, program::int, plan::char |
|
Students(id::int, name::text, program::int, plan::char [6]) |
Under tests/, you will find three subdirectories:
0_sanity-checks
Doesn't contain data to be stored. Assumes that you have an (empty) database with a PersonName data type loaded, and allows you to run a bunch of simple checks on values of type PersonName. Useful for testing your parsing function pname_in(), your output function pname_out and your operators.
1_users
Contains a schema and database for a table with just one attribute of type
PersonName. There are three data files of differing sizes. You should create a new database, add the schema and then insert the data from one data file.
There are a series of tests to check whether all of the data was loaded and can be manipulted correctly.
2_students
Contains a schema for a table of student data. There are four data files ranging in size from 100 tuples to 100000 tuples; each data file is a superset of the preceding data file. The queries test a wide range of PersonName functionality, including indexing. You should definitely check whether your code can deal with data4.sql; some bugs don't manifest themselves until you try with large amounts of data.
You can look at the tests in these directories and copy-paste them into a psql session if you want to perform individual tests. Or you can run a comprehensive set of tests using the Python script described below. Each directory contains an info.txt file giving more details about the files in the directory.
Next, you should move your pname.c and pname.source files into the new testing directory.
|
... this is the probable location that your files are in ... ... if they are somewhere else copy from that location instead ... $ cp postgresql-15 .6/s rc/tutorial/pname .c testing/pname .c $ cp postgresql-15 .6/s rc/tutorial/pname .source testing/pname . |
From this point on, you can do your assignment development in the testing/ directory. The provided Makefile will do the same job as the one in the postgresql-15.6/s rc/tutorial/ directory. Don't forget to run the command:
|
$ source /localstorage/$USER/env |
and (re)start your PostgreSQL server every time you want to do a development/debugging session on your PersonName data type.
You can now run the tests.
|
$ ssh nw-syd-vxdb $ source /localstorage/$USER/env $ p1 $ cd /localstorage/$USER/testing $ ./run_test .py ... internal logs from postgresql a re placed in pg.log ... ... external logs from the testing script a re placed in test.log .. |
|
... start by reading test.log after run_test.py has finished ... |
You can download and run the tests at home with this link (WARNING: contains a very large file). We won't provide assistance in setting up your home environment for doing this assignment; it's your problem to work it out.
One thing you will need to do is edit run_tests.py and remove the check that you are on vxdb at the start of main(). And change the check for a postgresql-15.6 directory to suit local conditions.
NOTES:
. pname.source should do the bare minimum needed to create the
PersonName type as specified.
pname.source should not try to create any tables, insert any data, or drop the PersonName type.
. Everything done in pname.source should be undo-able with "DROP TYPE PersonName CASCADE".
(but pname.source should not actually do this ... see previous point)
These tests remove some information like timing, paths, and width from output to allow for comparison to expected output.
Do your own tests, especially to make sure that queries are completed in a reasonable time and you are storing data in a reasonable amount of space.
Please report any issues ASAP on the Webcms3 comments for this page.
Have fun, jas