// Copyright (c) 2006 IBM

ICU Tech Meeting
02/17/2006 2PM
Andy
George
Ram
Mark
Deborah
Vladimir
Markus
Tex
Doug
Eric
Agenda:
* dictionary break iterator
* wrap up on Freezable UText
* ICU utf-8 usage discussion
* charset recognition C port postmortem
* discuss! bugs
* Mark/Yoshito - globalization prefs
* Mark/Flexible date/time stuff

* Deborah (dictionary break iterator):
- Break iterator dictionary - to have UData header or not
- we want dictionary binary data to be in a bundle
- where udata should go. Writing data will be a ures call. 
- Dictionary C++ class doesn't need to know about udata info
- loading code needs to know about udata. But that code is part of break engine factory. Factory will load the dictionary and give it to the dictionary engine. 
* wrap up on Freezable UText
- andy questions: name of the function - mark: we settled on 'freeze' & 'freezable'
- break iterator already has isWritable and readOnly parameter.
- UText should have isReadOnly. 
- Only for UText, the function should be setReadOnly. isReadOnly and readOnly as a param on clone. 
- Deborah's bug requested that characters could cross chunk boundaries (4873). Markus strongly disagrees, this makes a mess on macros. Motivation: data structures may not look at the contents of chunks.
- Deborah to send out before and after changes for ICU code. 
* ICU utf-8 usage discussion
- Tex: people are asking for functionality that is in ICU, but with utf-8 interface. 
- Andy: general solution is UText - we want to expand the list of ICU services that accept UText and UText wrappers for utf-8. What are services with high importance. 
- Tex: timeframe? 
- Andy: UText exists on break iteration. regex is a high priority.
- Mark: Google has the same interest. On the C side, utf-8 is standard. 
- Apple: break iteration, regex
- Tex: constrain is that many technologies are utf-8 based. It is not just light processing.
+-+- progress on standing code contribution document. -+-+
- Deborah: Andy was planning to convert break iterator to UText. Yes, it is going to happen in near future.
* Mark/Yoshito - globalization prefs - needs CLDR data
* Mark/Flexible date/time stuff. Mark working on J version. Deborah working on C version. 
- Tex: date/time format perfomance improvements? 
- Deborah: share read only calendars.
+ perfomance agenda item for next 
* charset recognition C port postmortem
* discuss! bugs
+ Tex: support for additional encodings. Big5 was added. Found a lot of big5 mislabeling. Everything is therefore made Big5 HKSCS.