RFR: JDK-8073152: Update Standard/ExtendedCharsets to work with module system
Alan.Bateman at oracle.com
Mon Feb 16 19:55:15 UTC 2015
On 16/02/2015 18:46, Xueming Shen wrote:
>> 1. Is list_old needed, I can't tell if this is checked in for use by
>> future archaeologists.
> No, it's not used by any code. I just added it the last minute to be a
> reference for the future, as
> I'm deleting the original list sbcs, dbcs and extsbcs.
Okay, I'm just thinking of someone finding this file in 6 months or 6
years and wondering whether it is used in the build or not.
>> 2. Hasher.java is showing up in the webrev as a new file, was this
>> build.tools.hasher.Hasher and so we know have two copies?
> "This" Hasher.java has a different use interface and works on a pair
> of key/value lists directly (instead
> of parsing the key/value from a std in). Given the implementation is
> simple/small enough, I just copied/
> pasted the parts I need from Mark's code, left the original one
> untouched (I kinda remember it is used
> by other as well)
Okay, just trying to reduce duplication.
>> 4. Are there any tests that need to be updated?
> Only NIOCharsetAvailabilityTest.java. But I believe that test should
> not stop working for a while as
> it tried to "parse" the rt.jar and charsets.jar to collect the charset
> class names, which obviously
> does not work in module. It need to be rewritten with the new
> iterating mechanism. (it is supposed
> to be the last file in webrev. But it appears it was not there in
> first webrev)
Thanks, I was wondering if any tests were directly referencing charsets
that moved on some platforms.
> Phil is moving away from using those sun.nio.cs.ext directly. My
> change in motif should have no conflict
> with his changes, in fact they probably no longer needed. Just be
> there in case I will push first :-)
I think this looks okay.
More information about the build-dev