Please review: 7029048: LD_LIBRARY_PATH launcher changes

Kumar Srinivasan kumar.x.srinivasan at oracle.COM
Tue Apr 5 23:44:49 UTC 2011


> This looks okay to me and the change in java_md.c is mostly bringing 
> back what you purged.
> I briefly compared this fix with that changeset and a lot of testing 
> to verify this fix.  Minor comment:
>    java_md.c a typo in L128 "propogate" - should be "propagate"
oops will do.

>    test/tools/launcher/ should it be marked with 
> "othervm" test in @run clause?

No. not really,  since the "driver/main" creates "test vectors" and 
compares them, ie.
it configures the inputs, reaps the output, and analyzes the results, 
therefore this
can remain in the samevm.



> Mandy
> On 04/04/11 16:36, Kumar Srinivasan wrote:
>> Hello core-libs friends,
>> Background:
>> Earlier in the jdk7 development  setting of LD_LIBRARY_PATH was purged
>> in the launcher, this is a good thing!, Joe has an excellent blog 
>> which explains
>> the rational etc. etc. here:
>> More background:
>> Unfortunately!, the library versioning  that Kelly proposed earlier 
>> on build-dev,
>> <attached here for reference >  is infeasible  for jdk7, though this 
>> is the correct
>> approach.
>> Thus the  approach contained here, allows us to provide a 
>> transitional path to the
>> version scheme in the future. This is also the reason, the c-code is 
>> encapsulated
>> within a conditional and the test has been separated, such that these 
>> can be jettisoned
>> when required.
>> This explains the current proposed fix:
>> The old changeset for reference:
>> The old changeset:   
>> and finally here is the proposed webrev:
>> Thanks
>> Kumar

More information about the core-libs-dev mailing list