Tested 3-23-99 Win32 build. In the above URL page, the test case #3 has Hankaku "\" and the test case #16 has Zenkaku "��". After both "\" and "��" are Zenkaku numbers. They do not break lines even though you move mouse to change the window size. They should break lines.

We should break between number followed by a "\", but not "\" followed by number. Currently, it does not break in neither places.

Fix it and check in today. The case #16 should be fine now. The #3 should be correct if we really convert the Shift JIS into Yen sign. It currently does not break because it is converted into U+005C. (It work if we use ¥). The #3 problem now is only the conversion issue reported as 4238

I tested this in 2000-09-21-08 Win32, 2000-09-21-04 Mac, and 2000-09-21-10 Linux build. This does not work.

unsetting target milestone, M4 has been out for a long time.

I think this is regression. Reassign to shanjian

This might be related with a change I made in line breaking. We are using dominating space if there is no CJK character involved. And because unicode is used for internal processing, zenkaku character might not be necessarily considered as CJK character. It is well understood that no Word/Line break algorithm is perfect, and I believe the above behavior is compromisable.

This works fine in 4.x. This follows the Japanese kinsoku rule. I reopen this.

explaination about the 2 changes: 1, if a ZenKaku character is met, we go to CJK route. Unicode code point in range of ff00 to ffff is added. (Yes, I know some code points in this range is not yet or will never be allocated.) 2, In CJK route, 0x5c restore back to its category according to 4501.

sr=ftang

sr=erik

fix checked in.

Verified in 2001021204 Mtrunk build.