Previous | Next
Topic: Setup and Teardown procedures
Conf: 538, Msg: 39073
From: Steven Deneir (firstname.lastname@example.org)
Date: 10/30/00 01:47 AM
Setup and Teardown procedures Steven Deneir sdeneir email@example.com
utPLSQL is a great initiative !
Still some questions / remarks
Apparently the setup and teardown procedures are launched only at the beginning and at the end of the unit-test.
This involves some problems : if the unit to test reads and manipulates data, the tester must take these changes into consideration for the following test, which isn't that easy.
Some people will argue to use the Rollback statement. But what with functions that issue a commit statement and that are used in the unit that must be tested ?
It would be nice to have the possibility to specify for each test whether or not the setup/teardown should be re-executed. In that way there is no performance loss when no data-manipulation takes place.
Most of the time there are more procedures/functions to test within a package. But I only see 1 setup/teardown procedure. Does that mean that for all the routines to test in the package there is only 1 setup/teardown procedure ?
If this is the case, the tester must make sure that all necessary data is available for all procedures that will be tested...
Maybe there should be a setup/teardown for each procedure/function within a package.
In the doc 'Build Test Packages.htm' there is a line indicating :
exec utgen.testpkg ('str', output_type_in => utplsql.c_file)
But utplsql.c_file does not exists, I think this should be :
exec utgen.testpkg ('str', output_type_in => utgen.c_file);
Is it possible not to use scripts (OS-files) where the source is stored ? I am always working with one or another tool, so I rarely use scripts.
This goes for the recompilation of the packages and for the generation of the framework.