> but I'm optimistic because in most other parts of > OpenJDK, shell script tests do not set TMP. I normally only set the variables when the test need it explicitly. So I'm also optimistic about your changes.
Looks fine to me. Xuelei On 8/30/2011 3:34 PM, Weijun Wang wrote: > Hi All > > 7083664: test hard code of using c:/temp but this dir might not exist > > Webrev is at -- > > http://cr.openjdk.java.net/~weijun/7083664/webrev.00/ > > Some of our regression tests set TMP variables on different platforms, > and on Windows, it's c:\temp. Unfortunately one of the test machines > does not have this directory and a call to zip fails. > > This fix simply removes all TMP setting lines in security-related tests. > Most are in sun/security/tools, with the exception of > > lib/security/java.policy/Ext_AllPolicy.sh > > My opinion is that touching TMP is simply a bad idea. A JPRT run also > shows it's not needed. > > I remember there was a time that a certain test harness using by the SQE > team stripped all existing environment variables, and caused troubles > when a test needed it. Little by little, the harness added some > variables and the tests were happy. I don't know if this code change > will break it, but I'm optimistic because in most other parts of > OpenJDK, shell script tests do not set TMP. Or maybe that test harness > (forget the name) is already not in use. > > Webrev for closed tests will go in another mail. > > Thanks > Max >
