Wiki-translation discussion

Forum List Topic List

[ prev topic ]  

How does the equivlancy rating account for generated content?

Posted by ricks99 on Fri 18 of Apr, 2008 16:20 GMT United States
How (or does) the wiki-translation module account for generated content (such as the {TOC} plugin for wiki pages)?

Consider the following example:

The actual page content for each is identical. However, because the German structure is incomplete, the generated TOC is not equivalent.

By default the translation module will show these two pages as identical — although from the user's point of view, they are clearly not.

(I was able to force the German page out-of-sync, by editing the English page, then saving the German page as a "partial translation".)

Are there plans to handled this type of generated content?
Reads: 62608
Link
Posted by Xavi on Mon 21 of Apr, 2008 07:40 GMT Catalan Countries
Hey, Rick, that's a good point!
+1 (it will be affecting most of doc.tw.o pages using structures....)
Link
Posted by alain_desilets on Mon 21 of Apr, 2008 13:05 GMT
> Hey, Rick, that's a good point!
> +1
> (it will be affecting most of doc.tw.o pages using structures....)

Can someone explain the issue in more detail?

I'm kind of new to Tiki, so I still don't know all of its intricacies.

Alain
Link
Posted by ricks99 (not registered) on Mon 21 of Apr, 2008 13:27 GMT
> > Hey, Rick, that's a good point!
> > +1
> > (it will be affecting most of doc.tw.o pages using structures....)
>
> Can someone explain the issue in more detail?
>
> I'm kind of new to Tiki, so I still don't know all of its intricacies.
>
> Alain

Alain:

Consider the two pages I referenced above.

The actual wiki text for the English page is:
!TikiWiki for {TAG(tag=>strike)}Dummies{TAG} ''Smarties''
A beginner's guide to using TikiWiki CMS/Groupware.

!!!!Table of Contents
{toc}


The actual wiki text for the German page is:

!TikiWiki für {TAG(tag=>strike)}Dummies{TAG} ''Pfiffige''
Anfängerleitfaden für TikiWiki CMS/Groupware.

!!!!In diesem Abschnitt
{toc}


Tiki will see these pages as "identical" — the actual text is fully translated from English-German.

But... the toc is a special plugin that automatically generates the table of contents, based on the other pages in the structure. In this case, the English structure is complete, but the German structure is not.

Therefore, an end-user who accesses both pages will see very different content on each — the English page is much more complete than the German page.

I've taken a look at some of the translation code, and it appears (based on my limited knowledge) that the text comparison is being done against the data column for each page. But this column does not include the generated content (in this case, the TOC listing).

I was going to suggest using the cache column instead (which does include the content) — but what about sites that don't use wiki page caching?

Perhaps this is coupled with the discussion of translating structures?

-R
Link
Posted by alain_desilets on Mon 21 of Apr, 2008 13:53 GMT
Got it rick. Thx for your patient spelling out of the issue.

I have a feeling that indeed, this will tie in with the issue of translating structures.

I will add a tracker item for this to the project's todo list.

What priority would you say this is in the context of Tiki for Smarties? Low, Medium, High?

Alain

BTW: I will be working on CLWE stuff all day Tuesday. Are you available for a Skype chat at some time during the day? I'm based in Ottawa (EST time).

> > > Hey, Rick, that's a good point!
> > > +1
> > > (it will be affecting most of doc.tw.o pages using structures....)
> >
> > Can someone explain the issue in more detail?
> >
> > I'm kind of new to Tiki, so I still don't know all of its intricacies.
> >
> > Alain
>
> Alain:
>
> Consider the two pages I referenced above.
>
> The actual wiki text for the English page is:
>
!TikiWiki for {TAG(tag=>strike)}Dummies{TAG} ''Smarties''
> A beginner's guide to using TikiWiki CMS/Groupware.
>
> !!!!Table of Contents
> {toc}
>

>
> The actual wiki text for the German page is:
>

> !TikiWiki für {TAG(tag=>strike)}Dummies{TAG} ''Pfiffige''
> Anfängerleitfaden für TikiWiki CMS/Groupware.
>
> !!!!In diesem Abschnitt
> {toc}
>

>
> Tiki will see these pages as "identical" — the actual text is fully translated from English-German.
>
> But... the toc is a special plugin that automatically generates the table of contents, based on the other pages in the structure. In this case, the English structure is complete, but the German structure is not.
>
> Therefore, an end-user who accesses both pages will see very different content on each — the English page is much more complete than the German page.
>
> I've taken a look at some of the translation code, and it appears (based on my limited knowledge) that the text comparison is being done against the data column for each page. But this column does not include the generated content (in this case, the TOC listing).
>
> I was going to suggest using the cache column instead (which does include the content) — but what about sites that don't use wiki page caching?
>
> Perhaps this is coupled with the discussion of translating structures?
>
> -R
Link

Posted by Anonymous on Mon 21 of Apr, 2008 14:01 GMT
At this time, you can activate the multilingual structures and use a single structure for all languages rather than one structure per language. The multilingual structures will read the user's preferences to display the pages in the list in the best possible language. In your case, most pages would be displayed in English until they are translated to German.

At this time, I have been made aware that there are performance issues on very large structures. I still need to find some time to investigate the issue.
Link
Posted by ricks99 on Mon 21 of Apr, 2008 16:23 GMT United States
OK, I like (love) the idea of having a single structure. It would eliminate a lot of the maintenance issues that I face. But how will it handle automatically generated TOC elements?

For example, if my structure is:
  • Page A (available in EN, ES, and DE)
  • Page B (available in EN, ES)
  • Page C (available in EN, DE)
  • Page D (available in EN, ES)

What would the TOC generate for an EN visitor? For an ES visitor? For a DE visitor?

Ideally, I'd like to see it generate on-the-fly, based on the user's preferred language (or browser language for anonymous visitors). I'd also like to see it take into account the "better translation" notion as well, showing me better alternatives if available.

Am I asking for too much?

-R
Link

Posted by Anonymous on Mon 21 of Apr, 2008 16:50 GMT
Exactly what it does...

English would see all english.
German with secondary as english would get A/de B/en C/de D/en

Order is based on other languages you can read or browser settings as a fallback for anonymous users.
Link
Posted by ricks99 on Wed 23 of Apr, 2008 14:11 GMT United States
> Exactly what it does...
>
> English would see all english.
> German with secondary as english would get A/de B/en C/de D/en
>
> Order is based on other languages you can read or browser settings as a fallback for anonymous users.

But... one problem I've encountered is that the {toc} doesn't work for the pages that are not directly in the structure.

In my example, since only the EN pages are in the structure (the other pages are simply translations of the EN page), a German reader will see A-DE, but the {toc} on the page will be empty. :(
Link

Upcoming Events

No records to display