Localizing a TM1 model, part 2
Posted: Sun Mar 03, 2019 11:06 am
Hello all,
Edit: Topic is moving in the right direction, see 2nd post below for an update.
As you might know, I live in Belgium where we have 3 official languages: Dutch, French, German. Include English to the mix as well as TM1 models that are used globally and you understand that localizing a TM1 model is very important for the user experience AND maintaining/extending the model. So some time ago I started off here and in the meantime I made good progress.
We can for instance have this result in PAW:
whereby I localize using a TI process:
- visible in the screenshots: element names, dimension names
- not visible in the screenshots: cube name, subset names, view names, process names
- not looked at yet: hierarchies in PA speak
I can add the TI process for someone who wants to experiment as well.
We only need to set/change the language in the browser and refresh to see the update. If you want to update the UI in PAW in a different language, you need to close the book and reopen.
With the Caption alias this works fine (as documented in the manuals) but we can also localize any attribute.
For example, the month long name can be adjusted, but also the short name. The advantage is huge, instead of having six aliases (long name EN, long name FR, long name NL, short name EN, short name FR, short name NL) and even more with more languages/locales, we can suffice with "long name" and "short name".
Caption can also be a text attribute instead of an alias but my first concern is aliases.
Challenge
You will understand the one issue here. The users can switch aliases:
I would like to localize the 2 alias names too since someone who does not understand English (which is rare but still) will not want to see Caption and Caption_Short. Yes, we can train the users, and yes we can be consistent in our naming of aliases over the different dimensions/hierarchies but still I prefer to see:
- "Long name" and "Short name" if language = English
- "Nom long" and "Nom court" if language = French
- "Lange naam and "Korte naam" if language = Dutch
So the question is this:
- can we localize attribute names ? According to the documentation yes: https://www.ibm.com/support/knowledgece ... model.html
- in general, can we have in TM1 attributes on attributes ? (Control objects on control objects)
I agree that localizing models will be more of a challenge when users file an issue with screenshots in their own language, but to me being able to localize TM1 objects is a big step forward. The element names and translations will be probably stored in a relational database or text files as a second best approach.
I will test if the natural language search engine supports the localized names too such that users can write in their own language.
[Let's not forget that picklists towards subsets and dimensions might be impacted, so we (might) need to extract the principal element name.]
Thanks a lot !
Edit: Topic is moving in the right direction, see 2nd post below for an update.
As you might know, I live in Belgium where we have 3 official languages: Dutch, French, German. Include English to the mix as well as TM1 models that are used globally and you understand that localizing a TM1 model is very important for the user experience AND maintaining/extending the model. So some time ago I started off here and in the meantime I made good progress.
We can for instance have this result in PAW:
whereby I localize using a TI process:
- visible in the screenshots: element names, dimension names
- not visible in the screenshots: cube name, subset names, view names, process names
- not looked at yet: hierarchies in PA speak
I can add the TI process for someone who wants to experiment as well.
We only need to set/change the language in the browser and refresh to see the update. If you want to update the UI in PAW in a different language, you need to close the book and reopen.
With the Caption alias this works fine (as documented in the manuals) but we can also localize any attribute.
For example, the month long name can be adjusted, but also the short name. The advantage is huge, instead of having six aliases (long name EN, long name FR, long name NL, short name EN, short name FR, short name NL) and even more with more languages/locales, we can suffice with "long name" and "short name".
Caption can also be a text attribute instead of an alias but my first concern is aliases.
Challenge
You will understand the one issue here. The users can switch aliases:
I would like to localize the 2 alias names too since someone who does not understand English (which is rare but still) will not want to see Caption and Caption_Short. Yes, we can train the users, and yes we can be consistent in our naming of aliases over the different dimensions/hierarchies but still I prefer to see:
- "Long name" and "Short name" if language = English
- "Nom long" and "Nom court" if language = French
- "Lange naam and "Korte naam" if language = Dutch
So the question is this:
- can we localize attribute names ? According to the documentation yes: https://www.ibm.com/support/knowledgece ... model.html
- in general, can we have in TM1 attributes on attributes ? (Control objects on control objects)
I agree that localizing models will be more of a challenge when users file an issue with screenshots in their own language, but to me being able to localize TM1 objects is a big step forward. The element names and translations will be probably stored in a relational database or text files as a second best approach.
I will test if the natural language search engine supports the localized names too such that users can write in their own language.
[Let's not forget that picklists towards subsets and dimensions might be impacted, so we (might) need to extract the principal element name.]
Thanks a lot !