Re: Thoughts on bug 1254 (formerly 1174)

On Fri, Oct 17, 2014 at 5:46 AM, Pongrácz István <..hidden..> wrote:

Hi,

Regarding our pseudo-gettext files, I think it is a good format, at least one can use free softwares to translate it, even simple txt editor.

In the other hand, labels in general has some issues, due to that, English words usuall have different meanings in different context. Like 'to' means recipient in email context, date in time context etc. The other one is the 'account' and several other words.

In this case, it is hard to translate correctly to a specific language.

Please correct me, but when I translate a word, like 'to', I cannot manage that, where to use this translation.

I am sure prepositions are problematic there.

I saw other solutions, where almost every labels/texts has its own label (identifier) and the translator could translate like this:

contact_account_creditlimit_label="Credit limit" (in english file)

contact_account_creditlimit_label="Hitelkeres" (in the Hungarian file)

I also like the idea of some online systems, where the user in "dev mode" is able to rewrite/translate the label while he/she is using the system, clicking on a little icon.

In GNU Gettext you can use contexts to solve this problem. Have one string with different contexts, specify the desired context at translation time and away you go. Our modules do not support this however....

It's a shame because GNU Gettext is actually very mature. I am almost wondering if we need to start another GNU Gettext module since none of the ones I saw on CPAN seemed to work properly (maybe even use XS and create it as a language binding to the GNU utilities).

I know, it is not only about 1254, but translation infrastructure.

What do you think?

Cheers,

István

----------------eredeti üzenet-----------------
Feladó: "Chris Travers" <..hidden..>
Címzett: "Development discussion for LedgerSMB" <..hidden..>
Dátum: Thu, 16 Oct 2014 23:05:39 -0700
----------------------------------------------------------
Hi;

In the process of resolving various issues on bug 1174, issues with the Hungarian translation, it was found that Locale::Maketext::Lexicon and GNU gettext do not have compatible handling of %. As I have looked further into the code it looks like only a very narrow subset of gettext formatting functions are supported by our current libraries because parameterization is handled in fundamentally different ways.

GNU Gettext uses sprintf with appropriate qualifiers, and so % has to be escaped as %%. Locale::Maketext uses numeric identifiers with non-numeric ones being method names. What this means is that our .po files aren't really po files, but things that just kind of look like po files.

Initially we settled on gettext files because we wanted to have a standard format. However this is not what we have now. So the question is, going forward (1.5 and beyond) do we want to continue to use pseudo-gettext files? Or do we want to move to something else? If we side with the current status quo we certainly need to document it.

--

Best Wishes,
Chris Travers

Efficito: Hosted Accounting and ERP. Robust and Flexible. No vendor lock-in.

http://www.efficito.com/learn_more
------------------------------------------------------------------------------
Comprehensive Server Monitoring with Site24x7.
Monitor 10 servers for $9/Month.
Get alerted through email, SMS, voice calls or mobile push notifications.
Take corrective actions from your mobile device.
http://p.sf.net/sfu/Zoho
_______________________________________________
Ledger-smb-devel mailing list
..hidden..
https://lists.sourceforge.net/lists/listinfo/ledger-smb-devel

Best Wishes,

Chris Travers

Efficito: Hosted Accounting and ERP. Robust and Flexible. No vendor lock-in.

http://www.efficito.com/learn_more