On Saturday 31 Oct 2009 8:51:17 am Kartik Mistry wrote:

...

I also see a lot of words beginning with '_'. Is this meant to happen?

Same as 1st explanation when first character is used for shortcut. Example: _File.

thanks - that cleared some doubts. But I think the whole .po file should be manually reviewed. For example one string offered for translation is: <font color=xyz>

imagine what would happen if someone translated 'color'.

On Saturday 31 Oct 2009 11:45:00 am sankarshan wrote:

...

thanks - that cleared some doubts. But I think the whole .po file should be manually reviewed. For example one string offered for translation is:

<font color=xyz>

imagine what would happen if someone translated 'color'.

Strings that should not be translated can be appropriately marked in the file. This practice prevents stuff like - String, Float etc from being translated.

that is what I mean by manual review of the file. There are strings there that should not be translated - also strings that have been broken up and hence are untranslatable as the word order differs in Indian languages.

On Saturday 31 Oct 2009 12:05:30 pm sankarshan wrote:

On Sat, Oct 31, 2009 at 11:55 AM, Kenneth Gonsalves lawgon@au-kbc.org

wrote:

...

that is what I mean by manual review of the file. There are strings there that should not be translated - also strings that have been broken up and hence are untranslatable as the word order differs in Indian languages.

I apologize for the somewhat cryptic previous reply. Developers, when they are generating the .pot file, have the option of doing such marking. Doing a manual review, although feasible, becomes cumbersome with each release and, increase in number of strings for the file etc. In a similar manner, if certain terms are Trademarks etc, they can also be mentioned as a comment which makes the process of translation easy. So, even if someone does a rough cut translation via automated tools and, then decides to polish it up, the comments help a lot.

actually most Indian developers add i18n later, very few i18n their code by default. When 'bolting' i18n into their code, errors creep in which are only noticed by the translator. Once the coders get into the habit, then these problems will not occur. In this case, a manual audit by someone familiar with the code base is needed. We have a lot of duplicate strings: 'icq error', and 'ICQ error' for example. As well as the things I pointed out before. And I notice a hell of a lot of '<br>'s at the end of strings. It is a real pain to change the language and type those out. Poedit barfs if they are not included. <br> within a string is acceptable I suppose, but at the end of a string?

On Saturday 31 Oct 2009 12:59:42 pm sankarshan wrote:

...

'ICQ error' for example. As well as the things I pointed out before. And I notice a hell of a lot of '<br>'s at the end of strings. It is a real pain to change the language and type those out. Poedit barfs if they are not included. <br> within a string is acceptable I suppose, but at the end of a string?

You can copy the source english string across (there are shortcuts allowing that to happen in KBabel/lokalize) and, then replace the english within <br> with your local language translation. For me, that works out faster than switching inputs/languages to type the tags.

I tried localize, but could not find where to switch off automatic spellcheck - with automatic spellcheck, it is impossible to see what one is typing in indic. Poedit is cool - does spellcheck on english but not on indic. But you need ctrl-C and ctrl-V to cut and paste - that only works if the keyboard is english. But my point is simply this: why are <br>s at the end of strings put in the .po file? They should not be there.

On Monday 16 Nov 2009 1:23:55 pm Philip Tellis wrote:

...

actually most Indian developers add i18n later, very few i18n their code by default. When 'bolting' i18n into their code, errors creep in which are only noticed by the translator. Once the coders get into the habit, then these

FWIW, i18n was added into Everybuddy about 3 years after it was first developed and Ayttm inherited that code. The code then moved from developer to developer, not all of whom understand everything (or anything) about i18n. For the most part the project has always had 1-1.5 developers working on it unlike some larger projects that have 6-8 full time developers and tens of part time contributors. Given this, a manual review has always been impossible. One just takes what one gets and hopes someone sends in a patch or explains the problems clearly.

I understand the constraints you are working under - but my own mantra is that a coder who does not i18n'ise his code while he is writing it is unfit to write code. (this has been true for the last 5-6 years at least). Yet if I took a poll on this list - or on the Chennai LUG list where they claim to be more passionate about language, I would warrant that very few coders would put their hands up (if any at all). That said I appreciate the huge amount of headache adding i18n to an older app - I have done it a couple of times, and would hate to do it again.

On Sat, Oct 31, 2009 at 5:35 AM, Kenneth Gonsalves lawgon@au-kbc.org wrote:

what is 'Aycryption - status' is that meant to be Encryption - status?

"Aycryption" is "Ayttm Encryption", the name of a module. It's like "Torvalds Unix" and "Tux".

From: http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=229928