Short thoughts:
1. think about using PreHashedMap
2. think about using CharSequence build by StringBuilder as HashMap key.
This would save the char[] copies while instantiating the immutable
strings. More sophisticated you could use a simple byte[] containing the
ISO-8859 coded chars. Latter would speed up the the comparison, as only
half of raw bytes must be compared.
-Ulf
P.S.: I'm in vacation since Tuesday.
Am 26.08.2010 00:12, schrieb Martin Buchholz:
Ping - I haven't heard any news from you lately.
Is there anything holding up this change?
(I'm finally getting some time to work on jdk7 again...)
Martin
On Tue, Jun 29, 2010 at 06:21, Masayoshi Okutsu
<[email protected] <mailto:[email protected]>> wrote:
All the changes look good.
Now, I'll take care of CCC for the forName change.
Thanks,
Masayoshi
On 6/29/2010 8:54 AM, Martin Buchholz wrote:
On Mon, Jun 28, 2010 at 16:32, Masayoshi Okutsu
<[email protected]
<mailto:[email protected]>> wrote:
Hi Martin,
On 6/29/2010 6:18 AM, Martin Buchholz wrote:
Hi Masayoshi,
OK, I implemented your suggestion as a follow-on change
http://cr.openjdk.java.net/~martin/webrevs/openjdk7/UnicodeBlock2/
<http://cr.openjdk.java.net/%7Emartin/webrevs/openjdk7/UnicodeBlock2/>
I've just skimmed the webrev. The changes look good. I
will review them in
details later today. A question is, do we need CCC
approval on the forName
change? Strictly speaking, it's a behavioral change.
Yes, I am changing the javadoc. On the other hand, this
brings the
code and spec
into compliance with the Unicode standard, which was already
documented,
so one can regard this as a clarification and bug fix!
I leave it to you to decide whether to submit this to CCC.
Of course, it is an incompatible change, but the mildest such.
"''
Were you really depending on UnicodeBlock.forName("----basic __
latin") throwing?
"""
(I plan to commit the already reviewed earlier batch
of changes soon)
BTW: is jdk7 going to implement the 5.2.0 version of
the standard?
If so, then we should make some further changes, like
accepting "ASCII"
as an alias for "Basic Latin".
It's very likely Unicode 6.0 which should be released in
September. Yuka has
done all 5.2.0 changes, but she is waiting for some legal
approval. Oracle
legal takes much more time to review.
I see.... Legal response time was already very long during
the Sun era.
Martin
