PMC Meeting Minutes =================== 18 Feb 2004 10:00 AM PT Attendes ======== Deborah Goldsmith (Apple) Craig Cummings (Oracle) Markus Scherer George Rhoten Vladimir Steven R. Loomis Eric Mader Andy Heninger Alan Liu Agenda (provided by Markus) ====== Initial agenda items: - Technical discussions. I will try to prepare a list of Jitterbugs. - Also, Craig Cummings replied to me with: Note that I don't have any particular agenda items -- nor do I have any hidden agenda items like JSR-204. Or maybe I do, like what is ICU planning to add to make up for all the missing APIs that JSR-204 didn't implement? :( ---- list of Jitterbugs needing discussion ---- Jitterbugs tentatively scheduled for ICU 3.0 3505 .dat files should be bit-for-bit identical on all platforms 3373 does root need an entry for each display code? 3450 C++ Coding Guidelines, boilerplate 2825 Add new API to format/parse currency `Currency amount' support via NumberFormat API OR Formattable 2885 Currency API NumberFormat currency instance factory methods (for convenience) 3321 Standardize rule/pattern quoting RFE: Standardize rule/pattern quoting 2503 need a policy on UErrorCodes - reuse across services or per-service 2952 define and fix ParseError postContext handling 3386 locale explorer should use APIs 3451 RFE: Multi-Threaded Test Framework 3474 Serialization/Deserialisation of "com.ibm.icu.util.GregorianCalendar" incorrect investigate serialization for icu4j 3528 Fix Fragile Inheritance 3549 getLocale() revisited getLocale() rebooted 3607 RFE: Ensure consistent API parameter checking 2742 RFE: Perl Short form character names 3424 Perl aware UnicodeSets 2934 RFE: Thai-Latin transform in ICU4C Jitterbugs scheduled for after ICU 3.0 3504 Remove Endianness From Package Names 2027 RFE: consider charset name matching to ignore leading 0s 2893 currency formatting ugly with non-symbol currency names 3274 Korean Hangul rendering is incomplete : U+1100 Jamo support(with opentype fonts) is missing 2007 UnicodeSet contains(String) incorrect RFE: UnicodeSet.indexOf() 3095 C++ UnicodeSet iteration clumsy 3441 Extra UnicodeSet functions RFE: UnicodeSet.getNormalized() & UnicodeSet.toCase() 2903 RFE: convenience API for ID strings Unicode<->char * strings 2972 Don't check for NULL this or NULL references 2526 Style handling in transliterator 3464 hex escape transliterator needs to understand \xhh Minutes ======= 1. [Markus] Craig wants to talk about ICU Java support. We'll do that later in the meeting. 2. [Deborah] C API for regex for 3.0? [Andy] It will be done for 3.0 ("must do"). 3. #3505 [Eric] *.dat files work moving between platforms but they are not binary identical. They are not even the same size between Win32 and Linux. [Markus] EOL may be a difference. [Vladimir] We should do it so that we can know that files are correctly generated if they binary compare as equal between platforms. [Markus] Swapper tool should produce binarily identical file from another platform. [Alan] So the task is larger -- we actually want to canonicalize the data file format on all platforms, modulo the swapper tool. [Markus] We can probably fix this by taking care of line ends on differnt platforms so that they are the same. 4. #3373 [Markus] Does root need an entry for each display code? [Mark] Two goals are: #1. Be able to get a list for testing. #2. Be able to get a list for localizers. [George] Taking the example of Ram's tool, we don't want different tools implementing the locale lookup code. [Mark] Reason #3. For tools. Things we need a list for: - ISO country - ISO language - script - timezone - ISO currency - exceptional ISO3066 codes [Alan] If we had a small utility library that provided an API to iterate over these lists then test code, tools, localization utilities, etc., could all use this API. Then it doesn't matter to the user where the data is. [Andy] But we still might not want data to be spread out if someone wants to go look and see where the data actually is. [Markus] Let's move the lists out of root and create some mechanism to access these lists. 5. Java JSR 204. [Craig] What work do we have There were a number of things that JDK 204 did not implement and what is happening with them. [Mark] Things that are missing or that were not implemented can go into ICU4J, or in some cases are in ICU4J already. The other thing we will be doing is having alias methods that map ICU4J to Sun-like API. Most of what we'll be doing is adding aliases. 6. #3450 C++ coding guidelines. [Andy] Defer discussion until next meeting. 7. #2825 Currency parse/format [*] Conclusion is that we will add a new Object subclass on Java and a new Formattable flavor on C++ that contains a Number and a currency. The formattable flavor on C++ should not be constrained to a double -- it should be designed so that if/when we add bigdecimal support to ICU4C, it works. 8. #2885 Don't need new API now; refile last two lines as a separate bug. 9. #3321 Standardize rule/pattern quoting [Andy] We can't do anything without some initial research. [Mark] Two pieces. 1. Document what the conventions are. Do this in the 3.0 time frame, can be post-DCUT. 2. Actually make changes later. [Mark] Add on to the MessageFormat apostrophe bug to survey the other syntax uses. 10. [Markus] We will have ICU phone meetings every two weeks starting today. Please get your agenda items to me in advance of each meeting.