Hi,

Mon, 6 Dec 2010 11:40:24 +0200 (EET) Pekka Enberg <penb...@kernel.org>:

> Hi Ivan,
> 
> On Wed, 24 Nov 2010, Ivan Maidanski wrote:
> > Thank you merging these patches. I've attached the rest of my pending
> patches (33) - I posted them this Summer to classpath-bugs mailing list. (Of
> course, I don't know when and whether they ever would be committed to the
> GNU Classpath SVN.) The changelog entires accompanying the patches are in
> chlogs.txt.
> 
> I tested and merged this patch:
> 
> https://github.com/penberg/classpath/commit/f04218247e757318e61b7e497b2a7c2b1e0e02f6

Thank you.

> 
> I reverted this patch for now because I'm unable to reproduce the problem:
> 
> https://github.com/penberg/classpath/commit/6a1a037732fe5feebec34635546bef72c4c3e058

Sorry, which problem are you unable to reproduce?

The patch does:
1. improve speed (and reduces code size) slightly (but making some private 
methods static as well as making some private fields final);
2. fix comment typos;
3. in DecimalFormat.parse():
- always adjust pos and errorPos correctly on return;
- handle NaN and +-Infinity strings as in the RI (using startsWith instead of 
contains);
- speed up a little (by remove redundant startsWith calls);
- adjust stop position properly according to maximum decimal/fraction digits 
(if anybody thinks there should not be a stop position - we could discuss it 
while it exists it should be computed properly);
- work correctly with the custom-defined digits, 'e', plus/minus characters;
- catch double 'e' symbol (as well as presence of 'e' with exponental notation 
disabled);
- simplify the parse loop code (and reduce the code size slightly) by removing 
multiple occurrences of 'i--';
4. in SimpleDateFormat class: remove "public" for methods that are not a part 
of the public API;
5. in SimpleDateFormat.parse():
- optimize limit_digits case (and don't use pos parameter as a temp variable);
- don't test pos for null;
- use regionMatches(true) instead of toUpperCase().equals(s.toUpperCase());
6. in SimpleDateFormat.computeOffset():
- handle missing the first hour digit properly (instead of throwing NPE).

> 
> Pekka

Regards.

Reply via email to