Support for timestamped signed modules

[Mandy Chung]

  On 04/12/11 12:58, Sean Mullan wrote:
> New webrev: 
> http://cr.openjdk.java.net/~mullan/jigsaw/webrevs/timestamp/webrev.01/
>
ModuleFileFormat.java L796 - this commented line can be deleted.  Other 
than that, looks okay to me.

> I've made all of your suggested changes, except:
>
> > Both timestamp-test.sh and TimestampTest.java hardcodes the name of 
> the module
> > library, module name, and location. Would it be better to pass these 
> hardcoded
> > names from the shell test to TimestampTest as system properties or 
> input
> > arguments? These values might unlikely be changed but I would rather 
> avoid
> > having a hardcoded name in multiple files for the single test.
>
> These two tests are tightly coupled. The script is used to create the 
> keystore and other binaries (to avoid the "no binaries under source 
> control" rule) used by the Java program. I could change them all to 
> command-line arguments (there would be many) but it seems more 
> sensible for me to spend the time to eliminate the script entirely. 
> But that's a bit more work and I'll tackle that as an overall test 
> cleanup task a bit later.
>

Sounds good.

> > sun/security/pkcs/SignerInfo.java
> > L470: maybe assert tsa.length == 1, or it is guaranteed to be true??
>
> An assertion seems incorrect here, since this is verifying data sent 
> over the network by the server. I've added a check and it now throws 
> an exception if there is not exactly one signer.
>

Thanks
Mandy