Description of problem: Using scim as the input method, indic numeric characters cannot be successfully committed into each cell. After moving to another cell after entering the indic numeric characters, they gets converted automatically back to english numeric characters. Version-Release number of selected component (if applicable): openoffice.org-calc-2.0.1.1-11.2.2 scim-1.4.4-4.1 scim-m17n-0.2.0-1 m17n-db-bengali-1.3.2-2 How reproducible: Always Steps to Reproduce: 1.start oocalc with bn_IN locale, bn_IN.UTF-8 oocalc 2.select any cell and activate SCIM (ctrl-SPACE) 3.ensure bengali IME (M17N-bn-probhat) is selected from the scim manager found in systray 4.enter 12345 5. use -left or right arrow key to move to the next cell Actual results: indic numeric character changed to english numeric Expected results: indic numeric Additional info: if indic numeric characters are combine with indic characters, the conversion wont happen.
weird, it's *only* with calc, they *stick* as bengali in writer, and in the shared drawing layer code of drawing textboxes and so forth.
Ah yes, F11 for style list, right click on default, modify -> numbers select number category and fill in [NatNum1]0 for format code and calc "does the right thing", You'll see that dates and times are already in [NatNum1] mode. So this isn't a bug per-se, but instead an enhancement request to make the number default as [NatNum1]. Spreadsheets are tricky things.
ok, thinking through this there isn't any action that we should or could take. If you combine text and numbers you end up with text, and as such text is displayed as "format text", the same as was entered. Numbers only is a different beast. If you enter just numbers, either standard arabic numberals or bengali ones, or the numerals of *any* language, then they are auto-detected as numbers, converted to internal numbers and then displayed as "format number" with the number formatting code in use. i.e. all numbers are displayed the same way, whether entered using bengali numerals or arabic ones. It isn't possible to distinguish between 12345 or ১২৩৪৫ because they are the same thing to calc once detected as numbers. calc always uses "standard" as the default number formatting, which is arabic numbers in all locales. If a user wants to display in the "local" locale's number formatting he has to use number formatting "[NatNum1]0". Or if he wants to enter "text" that consists of only numbers, then he must use text formatting. Setting the default as [NatNum1]0 just for bengali doesn't make sense, it would have to be done for all languages, and doing means that entering literal arabic "12345" would automatically be detected as numbers and then displayed as the equivalent Bengali digits, so either way there is a scenario where the text entered is not the text displayed.