Apache OpenOffice (AOO) Bugzilla – Issue 76424
Date input in english automatically gets converted to bengali digits
Last modified: 2013-08-07 15:15:24 UTC
From Red Hat Bug #236529: Description of problem: Date typed in english in the Bengali India locale gets converted to bengali instead of remaining in english. Version-Release number of selected component (if applicable): openoffice.org-calc-2.0.4-5.4.17.1 How reproducible: Steps to Reproduce: 1. start OpenOffice Calc in Bengali India locale: LANG=bn_IN.UTF-8 oocalc & 2. Check that the bn keyboard is *not* selected as the input method 2. type in a date in the format 12-04-2007 3. press "Enter" and leave the cell Actual results: date typed in english automatically converts to bengali digits Expected results: date typed in english should remain in english Additional info: The digit conversion happens only for date formats. other numbers typed in english remain in english. date typed in bengali remains unchanged after leaving the cell. This behaviour is due to the [NatNum1] specifications as described here: http://qa.openoffice.org/issues/show_bug.cgi?id=47575 I am putting sayamindu dasgupta in the cc for this bug as he had modified the bn_IN.xml locale file to include this feature.
Hi, as the NatNum feature is set for the Format code, I re-assign this for evaluation to the BN Lang project. IMHO this is not a bug. Frank
I think we can remove the [NatNum1] tag from Bengali locale files. It seems bn_BD.xml is using bn_IN.xml's <LC_FORMAT>, so the patch is needed for bn_IN.xml Sayam, can you please submit a patch?
Removing [NatNum1] will not resolve the issue completely. Once the [NatNum1] is removed, date typed in bengali will automatically convert to English (as is the case of hi, gu, mr etc). Some additional feature has to be put in place so as to recognise the language of input and retain it. Manually setting [NatNum1] for each cell is not a feasible option either.
Created attachment 44822 [details] patch to remove NatNum1, if that's what you want to do
cmc, thanks for the patch. But we need solution for the issue mentioned by RunaB. I am not sure if that can be done.
I'm almost sure that it cannot be "fixed", but even if possible I question if it *should* be fixed as part of this issue. The situation is the same for every language which has "localized number digits", i.e. the rest of the indic languages and CJK ones. I would suggest that this issue restrict itself to what the default number format for Bengali should be. Either the same as the rest of the _IN languages and default to "English" digits or go it's own way and default to Bengali digits as is the current case.
Can this issue be raised as a Feature Request or a separate bug.
Currently Calc doesn't know anything about what native numerals lead to the recognition of number/date/time, it is all handled in lower layers of the number formatter's number recognition. In the number formatter there's also no mapping between input numerals and display formats. There's a default display format to use if no other format matching the input type was assigned to the cell, which in this case happens to have the [NatNum1] modifier, as do all predefined date/time format codes of the bn_IN locale. The bn_IN community now should decide what the preferred default display format should be: with or without native numerals, you can't have both. Display formats matching the input numerals would be a completely new feature.
Can the bn team decide a yes/no for 2.4 if we want to use "western" numerals by default for Bengali like other Indic languages or not.
Anyone available from the bengali localization team able to make decision as to which way to lean here ?
As there is no answer from the bengali team this task won't make it in time to OOo2.4 => retarget to 3.x
Is there anyone at the end of this bug ? I can apply the patch here to align bengali with all the other indic languages to get the same behaviour if you want ?
Ok, lets use "western" numerals by default for Bengali - though it is not the desired solution. So cmc's patch can be used now.
I'll look after this then
Bengali numerals treated the same as all other indic language numerals in cmcfixes44
Thanks, Caolan! Eike
cmc->er: Do you think you can verify that this is now the case in cmcfixes44 ?
Verified that all NatNum1 modifiers are removed from bn-IN format codes.
seen in DEV300_m13