Source string contextualization

[Nilamdyuti Goswami]

Hi all,

While translating some of the fedora packages we often come across some 
source strings whose context or meaning is not clear. This results in 
wrong translations which is discovered later while using the actual 
application. This in turn effects the concerned application.

To solve this issue, we have formed a Fedora SIG named "Source String 
Contextualizing Group" [1] aimed at
providing concise yet meaningful description of the concerned source 
strings in the source code itself to ensure the correctness and quality 
in the resulting translations.

We welcome feedback and suggestions.

Regards,
Nilamdyuti Goswami

[1]: https://fedoraproject.org/wiki/SSCG

[Kévin Raymond]

On Fri, Jan 24, 2014 at 1:22 PM, Nilamdyuti Goswami <ngoswami(a)redhat.com&gt; wrote:
>  Hi all,
>
>  While translating some of the fedora packages we often come across some
>  source strings whose context or meaning is not clear. This results in wrong
>  translations which is discovered later while using the actual application.
>  This in turn effects the concerned application.
>
>  To solve this issue, we have formed a Fedora SIG named "Source String
>  Contextualizing Group" [1] aimed at
>  providing concise yet meaningful description of the concerned source strings
>  in the source code itself to ensure the correctness and quality in the
>  resulting translations. 
It is *just* about improving the source code?
There is already a Transifex feature which can help provide context
without having to edit the source code.

Best would be able to proofread on the app itself..
See what is going on with GTK:
http://gil.badall.net/2013/01/23/i18n-dreams-eventually-come-true/
http://deckard.malizor.org/

that's amazing.



>  We welcome feedback and suggestions.
>
>  Regards,
>  Nilamdyuti Goswami
>
>  [1]: https://fedoraproject.org/wiki/SSCG
>
>
>  --
>  devel mailing list
>  devel(a)lists.fedoraproject.org
>  https://admin.fedoraproject.org/mailman/listinfo/devel
>  Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct 


-- 
Kévin Raymond
(shaiton)
GPG-Key: A5BCB3A2

[Nilamdyuti Goswami]

On 24-01-2014 07:37 অপৰাহ্ন, Kévin Raymond wrote:
>  On Fri, Jan 24, 2014 at 1:22 PM, Nilamdyuti Goswami
> <ngoswami(a)redhat.com&gt; wrote:
> > Hi all,
> >
> > While translating some of the fedora packages we often come across some
> > source strings whose context or meaning is not clear. This results in wrong
> > translations which is discovered later while using the actual application.
> > This in turn effects the concerned application.
> >
> > To solve this issue, we have formed a Fedora SIG named "Source String
> > Contextualizing Group" [1] aimed at
> > providing concise yet meaningful description of the concerned source strings
> > in the source code itself to ensure the correctness and quality in the
> > resulting translations.
>  It is *just* about improving the source code? 
It is about providing
Translators' comment or precise context to the 
confusing source strings in the source code of the application so that 
the resultant pot/po files have that context and it becomes easy for the 
translators to translate the confusing strings.
>  There is already a Transifex feature which can help provide context
>  without having to edit the source code. 
If Transifex has this feature, we would
like to know more about it 
because this can be a more direct solution to the problem. :-)
>  Best would be able to proofread on the app itself..
>  See what is going on with GTK:
>  http://gil.badall.net/2013/01/23/i18n-dreams-eventually-come-true/
>  http://deckard.malizor.org/
>
>  that's amazing. 
Yeah, we have tried *deckard* before and indeed it is amazing
but it is 
something which is used for proofreading. If a translator wasn't sure of 
a particular source string's meaning, did a wrong translation and 
eventually finds the mistake when he proofreads the actual UI in 
deckard, he has to do an extra work and make the correction in the 
respective po file whereas providing a proper context will ensure a 
correct translation in the first place. Proofreading the app, detecting 
the mistake, and then making the correction won't be required. :-)
> > We welcome feedback and suggestions.
> >
> > Regards,
> > Nilamdyuti Goswami
> >
> > [1]: https://fedoraproject.org/wiki/SSCG
> >
> >
> > --
> > devel mailing list
> > devel(a)lists.fedoraproject.org
> > https://admin.fedoraproject.org/mailman/listinfo/devel
> > Fedora Code of Conduct: http://fedoraproject.org/code-of-conduct
>
>

[Kévin Raymond]

>  It is *just* about improving the source code?
>
>  It is about providing Translators' comment or precise context to the
>  confusing source strings in the source code of the application so that the
>  resultant pot/po files have that context and it becomes easy for the
>  translators to translate the confusing strings. 
By doing that, I think you go to improve the way strings are defined also.
For example "a string" "could be spread" "in many" which
result on
something not translatable is some languages.

>  There is already a Transifex feature which can help provide context
>  without having to edit the source code.
>
>  If Transifex has this feature, we would like to know more about it because
>  this can be a more direct solution to the problem. :-) 
What seems to do the trick is
http://support.transifex.com/customer/portal/articles/1188235-comments-on...
one can comment a string (to all translators, but in the language he
choose), and event report an issue for his team or the developer (I
don't know if they listen to us...)

Of course, there is no perfect solution, but checking every project
sources might be a huge work. (with no end)

[Miloslav Trmač]

On Fri, Jan 24, 2014 at 3:07 PM, Kévin Raymond <shaiton(a)fedoraproject.org&gt;wrote:

>  On Fri, Jan 24, 2014 at 1:22 PM, Nilamdyuti Goswami
> <ngoswami(a)redhat.com&gt;
>  wrote:
>  > To solve this issue, we have formed a Fedora SIG named "Source String
>  > Contextualizing Group" [1] aimed at
>  > providing concise yet meaningful description of the concerned source
>  strings
>  > in the source code itself to ensure the correctness and quality in the
>  > resulting translations.
>
>  It is *just* about improving the source code?
>  There is already a Transifex feature which can help provide context
>  without having to edit the source code.
>  
Having translation context comments together with the source code is
beneficial for the same reason having API documentation together with the
source code is beneficial: they don't get out of sync.

Especially note that merely the English string is not always a sufficient
identifier of the context, unlike a function/method name in an API.  An
infamous case is the string "May" which may either be a month name or a
verb (in a software that is splicing sentences), and there already was an
application that used both.  Attaching a context within the translation
tool just wouldn't have worked.
     Mirek

[Ankit Patel]

On 01/24/2014 10:14 PM, Miloslav Trma wrote:
>  On Fri, Jan 24, 2014 at 3:07 PM, K�vin Raymond
>  <shaiton(a)fedoraproject.org <mailto:[e-mail]>> wrote:
>
>      On Fri, Jan 24, 2014 at 1:22 PM, Nilamdyuti Goswami
>      <ngoswami(a)redhat.com <mailto:[e-mail]>> wrote:
>      > To solve this issue, we have formed a Fedora SIG named "Source String
>      > Contextualizing Group" [1] aimed at
>      > providing concise yet meaningful description of the concerned source strings
>      > in the source code itself to ensure the correctness and quality in the
>      > resulting translations.
>
>      It is *just* about improving the source code?
>      There is already a Transifex feature which can help provide context
>      without having to edit the source code.
>
>
>  Having translation context comments together with the source code is
>  beneficial for the same reason having API documentation together with
>  the source code is beneficial: they don't get out of sync.
>
>  Especially note that merely the English string is not always a
>  sufficient identifier of the context, unlike a function/method name in
>  an API.� An infamous case is the string "May" which may either be a
>  month name or a verb (in a software that is splicing sentences), and
>  there already was an application that used both.� Attaching a context
>  within the translation tool just wouldn't have worked.
>  ���� Mirek
>
>  
+1

Absolutely,

The idea is to fix the problem from it's root rather than going the
reverse way. An ecosystem for the betterment of the content
representation is going to be helpful for
* developers
* translators
* end users
* and everyone else.

I can see that the SSCG SIG[1] is looking for an active support from
this larger group of developers to let them provide patches or updates
with regards to contextual information.

Thanks,

[1] https://fedoraproject.org/wiki/SSCG
-- 
Regards,
Ankit Patel
http://www.ankit644.com/

[Parag Nemade]

Hi Nilamdyuti,


On Fri, Jan 24, 2014 at 5:52 PM, Nilamdyuti Goswami <ngoswami(a)redhat.com&gt;wrote:

>  Hi all,
>
>  While translating some of the fedora packages we often come across some
>  source strings whose context or meaning is not clear. This results in wrong
>  translations which is discovered later while using the actual application.
>  This in turn effects the concerned application.
>
>  To solve this issue, we have formed a Fedora SIG named "Source String
>  Contextualizing Group" [1] aimed at
>  providing concise yet meaningful description of the concerned source
>  strings in the source code itself to ensure the correctness and quality in
>  the resulting translations.
>
>  
I see iok project have many locales available for translations in iok
package so maybe you people can help in improving source translation
strings by providing patches in bugzilla or upstream tracker. This is
really helpful to let translators understand the exact context of the
source string. This will help to have more accurate translations.

Regards,
Parag.

[Nilamdyuti Goswami]

On 27-01-2014 12:34 অপৰাহ্ন, Parag N(पराग़) wrote:
>  Hi Nilamdyuti,
>
>
>  On Fri, Jan 24, 2014 at 5:52 PM, Nilamdyuti Goswami 
>  <ngoswami(a)redhat.com <mailto:[e-mail]>> wrote:
>
>      Hi all,
>
>      While translating some of the fedora packages we often come across
>      some source strings whose context or meaning is not clear. This
>      results in wrong translations which is discovered later while
>      using the actual application. This in turn effects the concerned
>      application.
>
>      To solve this issue, we have formed a Fedora SIG named "Source
>      String Contextualizing Group" [1] aimed at
>      providing concise yet meaningful description of the concerned
>      source strings in the source code itself to ensure the correctness
>      and quality in the resulting translations.
>
>
>  I see iok project have many locales available for translations in iok 
>  package so maybe you people can help in improving source translation 
>  strings by providing patches in bugzilla or upstream tracker. This is 
>  really helpful to let translators understand the exact context of the 
>  source string. This will help to have more accurate translations.
>  Regards,
>  Parag. 
Thanks Parag for pointing out a package and providing your valuable 
suggestion. We shall work on the same.

Regards,
Nilamdyuti

On 24 January 2014 17:52, Nilamdyuti Goswami <ngoswami(a)redhat.com&gt; wrote:

>  Hi all,
>
>  While translating some of the fedora packages we often come across some
>  source strings whose context or meaning is not clear. This results in wrong
>  translations which is discovered later while using the actual application.
>  This in turn effects the concerned application.
>
>  To solve this issue, we have formed a Fedora SIG named "Source String
>  Contextualizing Group" [1] aimed at
>  providing concise yet meaningful description of the concerned source
>  strings in the source code itself to ensure the correctness and quality in
>  the resulting translations.
>
>  We welcome feedback and suggestions.
>  
I liked this.

I think this should become habit/protocol for translation community over a
period. Whenever one find any confusing string, they should report it back
to developer saying "we need context" Or they can give patch or suggestion
for same.

If every translator keeps on following it, in long term we will definitely
have term with good information.

Same time more information is not required for each and every word, only
for confusing terms. It make this task more achievable.

Regards,
Pravin Satpute

[Nilamdyuti Goswami]

On 27-01-2014 02:08 অপৰাহ্ন, pravin.d.s(a)gmail.com wrote:
>  On 24 January 2014 17:52, Nilamdyuti Goswami <ngoswami(a)redhat.com 
>  <mailto:[e-mail]>> wrote:
>
>      Hi all,
>
>      While translating some of the fedora packages we often come across
>      some source strings whose context or meaning is not clear. This
>      results in wrong translations which is discovered later while
>      using the actual application. This in turn effects the concerned
>      application.
>
>      To solve this issue, we have formed a Fedora SIG named "Source
>      String Contextualizing Group" [1] aimed at
>      providing concise yet meaningful description of the concerned
>      source strings in the source code itself to ensure the correctness
>      and quality in the resulting translations.
>
>      We welcome feedback and suggestions.
>
>
>  I liked this.
>
>  I think this should become habit/protocol for translation community 
>  over a period. Whenever one find any confusing string, they should 
>  report it back to developer saying "we need context" Or they can give 
>  patch or suggestion for same.
>
>  If every translator keeps on following it, in long term we will 
>  definitely have term with good information.
>
>  Same time more information is not required for each and every word, 
>  only for confusing terms. It make this task more achievable.
>
>  Regards,
>  Pravin Satpute 
Thanks Pravin. We hope it's implementation can help us yield
quality 
translations.

Regards,
Nilamdyuti

[Nilamdyuti Goswami]

----- Original Message -----
From: "Kévin Raymond" <shaiton(a)fedoraproject.org&gt;
To: ngoswami(a)redhat.com
Cc: "Development discussions related to Fedora"
<devel(a)lists.fedoraproject.org&gt;, "For participants of the Documentation
Project" <docs(a)lists.fedoraproject.org&gt;, "trans"
<trans(a)lists.fedoraproject.org&gt;, sscg(a)lists.fedoraproject.org
Sent: Friday, January 24, 2014 9:47:08 PM
Subject: Re: Source string contextualization

>  It is *just* about improving the source code?
>
>  It is about providing Translators' comment or precise context to the
>  confusing source strings in the source code of the application so that the
>  resultant pot/po files have that context and it becomes easy for the
>  translators to translate the confusing strings. 
By doing that, I think you go to improve the way strings are defined also.
For example "a string" "could be spread" "in many" which
result on
something not translatable is some languages.

Yes, in many cases the source string itself requires rephrasing, redefinition or
proper positioning (in case of split sentences). So, we have that aspect in our mind
as well.  We shall report the concerned developer or maintainer about such strings 
whenever we come across them and suggest the corrections. However, its implementation
depends entirely on the discretion of the developer or maintainer as unlike providing
contexts, it would be modifying the source string itself.

>  There is already a Transifex feature which can help provide context
>  without having to edit the source code.
>
>  If Transifex has this feature, we would like to know more about it because
>  this can be a more direct solution to the problem. :-) 
What seems to do the trick is
http://support.transifex.com/customer/portal/articles/1188235-comments-on...
one can comment a string (to all translators, but in the language he
choose), and event report an issue for his team or the developer (I
don't know if they listen to us...)

This is definitely a real time method of reporting an issue to the developer and is a good
feature
but to make it get noticed, I think filing a bug for the issue is better.

Of course, there is no perfect solution, but checking every project
sources might be a huge work. (with no end)

Agreed, but at least if a note is taken every time we come across confusing source strings
while translating,
later at the time of providing contexts, we don't need to scan the entire source code
line by line. Moreover,
making modification at the source code level is a more permanent solution to the problem.

[Dimitris Glezos]

On Thu, Jan 30, 2014 at 2:20 AM, Nilamdyuti Goswami <ngoswami(a)redhat.com&gt;wrote:

>  What seems to do the trick is
>
>  http://support.transifex.com/customer/portal/articles/1188235-comments-on...
>  one can comment a string (to all translators, but in the language he
>  choose), and event report an issue for his team or the developer (I
>  don't know if they listen to us...)
>
>  This is definitely a real time method of reporting an issue to the
>  developer and is a good feature
>  but to make it get noticed, I think filing a bug for the issue is better.
>
>  
The @mention in Transifex sends a notification. If you mark something as an
issue, the developer is notified automatically.

There are no plans to implement Bugzilla support. But, one can implement a
script outside of Transifex which takes an email and converts it to a
Bugzilla ticket quite easily.

-d


-- 
Dimitris Glezos
Founder & CEO, Transifex
https://www.transifex.com/