"none" is OK. But if you want a real JIT info the EM config file is the only
valid place to put it.
On 10/13/06, Alexei Zakharov <[EMAIL PROTECTED]> wrote:
Mikhail,
Yes, but I think we should have it by default. Another problem is that
we should return something even if we are in the interpreter mode.
Since this property is *mandatory* for all configs. I suggest using
something like "none" or etc.
Regards,
2006/10/13, Mikhail Fursov <[EMAIL PROTECTED]>:
> The most simple fix: put it into all EM configuration files if you want
it
> to depend on JIT configuration is used.
>
>
> On 10/13/06, Alexei Zakharov <[EMAIL PROTECTED]> wrote:
> >
> > Hi all,
> >
> > I found that we have some problems with system properties in DRLVM. In
> > the javadoc for java.lang.System class the particular list of
> > mandatory properties that should be returned by the getProperties()
> > method is described. However, DRLVM does not provide the value for the
> > mandatory "java.compiler" property. This leads to some problems
> > including the test failure from the java.lang.management test suite
> > (HY-1407). This problem was mentioned by Andrey Yakushev [1]. Below
> > is the small test that checks all mandatory properties:
> >
> > <--
> > import java.util.*;
> >
> > public class TestVMProps {
> >
> > static String[] mandatoryProps = new String[] {
> > "java.version",
> > "java.vendor",
> > "java.vendor.url",
> > "java.home",
> > "java.vm.specification.version",
> > "java.vm.specification.vendor",
> > "java.vm.specification.name",
> > "java.vm.version",
> > "java.vm.vendor",
> > "java.vm.name",
> > "java.specification.version",
> > "java.specification.vendor",
> > "java.specification.name",
> > "java.class.version",
> > "java.class.path",
> > "java.library.path",
> > "java.io.tmpdir",
> > "java.compiler",
> > "java.ext.dirs",
> > "os.name",
> > "os.arch",
> > "os.version",
> > "file.separator",
> > "path.separator",
> > "line.separator",
> > "user.name",
> > "user.home",
> > "user.dir"
> > };
> >
> > public static void main(String argv[]) {
> > Vector<String> missed = new Vector<String>();
> > Properties props = System.getProperties();
> >
> > for (String mandProp: mandatoryProps) {
> > boolean found = false;
> > Enumeration<String> propEnum = (Enumeration<String>)
> > props.propertyNames();
> >
> > while (propEnum.hasMoreElements()) {
> > String prop = propEnum.nextElement();
> >
> > if (mandProp.equals(prop)) {
> > found = true;
> > break;
> > }
> > }
> >
> > if (!found) {
> > missed.add(mandProp);
> > }
> > }
> >
> > System.out.println("Number of missed props: " + missed.size
());
> > System.out.println("Missed props: " + missed);
> > }
> > }
> > <--
> >
> > If invoked on DRLVM:
> >
C:\mydoc\projects\tests>C:\Java\harmony-hdk-r450941\jdk\jre\bin\java.exe
> > TestVMProps
> > Number of missed props: 1
> > Missed props: [java.compiler]
> >
> > On IBM VME:
> > C:\mydoc\projects\tests>C:\Java\harmony-bin\bin\java.exe TestVMProps
> > Number of missed props: 0
> > Missed props: []
> >
> > On RI:
> > C:\mydoc\projects\tests>C:\Java\jdk1.5.0\bin\java.exe TestVMProps
> > Number of missed props: 1
> > Missed props: [java.compiler]
> >
> > You may notice that RI also does not provide the value for
> > "java.compiler". BTW, we have a test in the DRLVM test suite that
> > should check all mandatory properties –
> > java.lang.SystemExtensionTest.test_getProperty_Str(). But it seems the
> > author of the test makes a muddle of logical operators – puts OR
> > instead of AND. So the test now checks if at least one of mandatory
> > property is set.
> >
> > IMHO we should return something for the "java.compiler" property name.
> > Any thoughts?
> >
> > [1]
> >
http://mail-archives.apache.org/mod_mbox/incubator-harmony-dev/200610.mbox/[EMAIL
PROTECTED]
> > (problem 2)
--
Alexei Zakharov,
Intel Enterprise Solution Software Division
---------------------------------------------------------------------
Terms of use : http://incubator.apache.org/harmony/mailing.html
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
--
Mikhail Fursov
