NetBeans IDE 6.7 RC3 (Build 200906142201)
Java: 1.6.0_07; Java HotSpot(TM) 64-Bit Server VM 1.6.0_07-b06-57
Mac OS X 10.5.7; SJIS; ja_JP (nb)
I got this feedback from Japanese community.
I'll attach the sample project using UTF-8 as project encoding.
Try to search e.g. "main" in the project. It should find strings
from both Main.java and Main1.java. However, it fails for Main.java.
The difference is, in Main.java, 4096th character in line 38 is Japanese.
It looks a regression. It does not happen on 6.5.1.
I understand many developers are using multibytes in their programs.
This basic functionality should not be broken for 6.7 FCS. Please evaluate.
Created attachment 83607 [details]
Fixed in the trunk by rejection of using FileEncodingQuery.ProxyCharset.ProxyDecoder which uses buffer of 4096 bytes to
decode. Now we are decoding whole file at once without using any buffers, so no situations with part of the multibyte
character should appears.
Integrated into 'main-golden', will be available in build *200906191401* on http://bits.netbeans.org/dev/nightly/ (upload may still be in progress)
User: Andrey Yamkovoy <firstname.lastname@example.org>
Log: Fix for #167137 - I18N : [regression] search does not work if a multibyte exists at 4096th bytes
Thank you! It's great!
I've verified it on NetBeans IDE Dev (Build 200906191401).
The fix has been ported into the release67_fixes repository.
Product Version: NetBeans IDE 6.7.1 RC (Build 200907150227)
Java: 1.6.0_14; Java HotSpot(TM) Client VM 14.0-b16
System: Windows XP version 5.1 running on x86; Cp1250; cs_CZ (nb)
Verified in 6.7.1 RC
Search results now contains both - Main and Main1. Among others :)