On 29/08/2014 22:07, Naoto Sato wrote:
I incorporated the suggestions from Mandy and Alan. Also one change
since the last webrev was to revert the hard-coding of the supported
locales list back to the one which dynamically generates the lists at
the build time. I initially thought static listing of locales would be
less complex as to the build script/makefile, but on second thought
it's less evil than possible future regressions.
Please review:
http://cr.openjdk.java.net/~naoto/8038436/webrev.5/
I think this looks okay and I assume you'll create another issue to
re-examine the error handling as I do think that does need to be looked
at again (my main concern is that it will silently failover and that
could be very hard to diagnose).
One comment on the test Bug8038436 is that it sets java.ext.dirs on the
assumption that the locale and cldr data wouldn't be found. I suspect
this will need to be re-worked once we get to the point of linking
modules into the image. It may be that the ext directory is empty by
default (localdata.jar and cldrdata.jar will not exist).
-Alan.
