Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Test Pilot locale support / detection
8 views
Skip to first unread message
Eduardo Trápani
Jul 24, 2017, 2:46:29 PM7/24/17
to dev-l10n
Hi,
I tried to access snooze-tabs
(https://testpilot.firefox.com/experiments/snooze-tabs) and got this
message:
Just a couple of things:
* I find the text misleading. As far as I know, the experiment it is
simply not translated into that language, meaning that things will work
just the same and I can test it if I understand another language to
which it has been translated. So "not supported" should be more like
"not translated/not localized".
And yes, I think it makes a big difference because users expect their
localized versions to do the same as the "main" (en-US) version, to be
full versions of the product. They know some parts might not be
translated, but then they can choose another language, or live with
en-US (Web Developer, SUMO, ...). Imagine if SUMO had such a notice.
The current message is kind of hostile to those locales that did not
translate Test Pilot because it says that the product is not supported
(as if in "it might not work"), leaving the impression of a
not-so-capable, crippled browser, because of the localization.
* I see no way to discover if there is another language I understand, to
which the experiment has been translated. But I don't know if it would
be technically possible to use a localization to language that is not
that of the browser.
* "My language" should probably be one in the list of languages in
Accept-Language. But it's not, because I also have en-US. And even if I
leave only en, en-US I receive the same message.
Eduardo.
I tried to access snooze-tabs
(https://testpilot.firefox.com/experiments/snooze-tabs) and got this
message:
Just a couple of things:
* I find the text misleading. As far as I know, the experiment it is
simply not translated into that language, meaning that things will work
just the same and I can test it if I understand another language to
which it has been translated. So "not supported" should be more like
"not translated/not localized".
And yes, I think it makes a big difference because users expect their
localized versions to do the same as the "main" (en-US) version, to be
full versions of the product. They know some parts might not be
translated, but then they can choose another language, or live with
en-US (Web Developer, SUMO, ...). Imagine if SUMO had such a notice.
The current message is kind of hostile to those locales that did not
translate Test Pilot because it says that the product is not supported
(as if in "it might not work"), leaving the impression of a
not-so-capable, crippled browser, because of the localization.
* I see no way to discover if there is another language I understand, to
which the experiment has been translated. But I don't know if it would
be technically possible to use a localization to language that is not
that of the browser.
* "My language" should probably be one in the list of languages in
Accept-Language. But it's not, because I also have en-US. And even if I
leave only en, en-US I receive the same message.
Eduardo.
Francesco Lodolo
Jul 24, 2017, 3:11:02 PM7/24/17
to dev-l10n
2017-07-24 20:46 GMT+02:00 Eduardo Trápani <etra...@gmail.com>:
> Hi,
>
> I tried to access snooze-tabs
> (https://testpilot.firefox.com/experiments/snooze-tabs) and got this
> message:
Error message is not visible, but I assume you mean: "This experiment
> Hi,
>
> I tried to access snooze-tabs
> (https://testpilot.firefox.com/experiments/snooze-tabs) and got this
> message:
is not supported in your language (eo)."
> Just a couple of things:
>
> * I find the text misleading. As far as I know, the experiment it is
> simply not translated into that language, meaning that things will work
> just the same and I can test it if I understand another language to
> which it has been translated. So "not supported" should be more like
> "not translated/not localized".
"available" or "translated" would be less open to confusion.
Can you please file an issue here?
https://github.com/mozilla/testpilot/issues
> * I see no way to discover if there is another language I understand, to
> which the experiment has been translated. But I don't know if it would
> be technically possible to use a localization to language that is not
> that of the browser.
>
> * "My language" should probably be one in the list of languages in
> Accept-Language. But it's not, because I also have en-US. And even if I
> leave only en, en-US I receive the same message.
or en as first language
There's no way to switch the locale, I filed an issue a while ago but
it was closed without a solution (just realized)
https://github.com/mozilla/testpilot/issues/2271
Francesco
Michael Wolf
Jul 24, 2017, 3:54:04 PM7/24/17
to
Eduardo Trápani schrieb:
I checked the source code of the page. There is a meta tag in the <head>
area for available languages that does not contain "eo". Maybe this is
the reason why Esperanto does not work:
<meta name="availableLanguages"
content="en-US,cs,de,dsb,es-ES,es-MX,fr,fy-NL,hsb,hu,it,ja,kab,nl,nn-NO,pt-BR,pt-PT,ru,sk,sl,sq,sr,sv-SE,uk,zh-CN,zh-TW"/>
Regards,
Michael
> * I see no way to discover if there is another language I understand, to
> which the experiment has been translated. But I don't know if it would
> be technically possible to use a localization to language that is not
> that of the browser.
>
> * "My language" should probably be one in the list of languages in
> Accept-Language. But it's not, because I also have en-US. And even if I
> leave only en, en-US I receive the same message.
Hi Eduardo, hi Francesco,
> which the experiment has been translated. But I don't know if it would
> be technically possible to use a localization to language that is not
> that of the browser.
>
> * "My language" should probably be one in the list of languages in
> Accept-Language. But it's not, because I also have en-US. And even if I
> leave only en, en-US I receive the same message.
I checked the source code of the page. There is a meta tag in the <head>
area for available languages that does not contain "eo". Maybe this is
the reason why Esperanto does not work:
<meta name="availableLanguages"
content="en-US,cs,de,dsb,es-ES,es-MX,fr,fy-NL,hsb,hu,it,ja,kab,nl,nn-NO,pt-BR,pt-PT,ru,sk,sl,sq,sr,sv-SE,uk,zh-CN,zh-TW"/>
Regards,
Michael
Eduardo Trápani
Jul 24, 2017, 4:04:58 PM7/24/17
to dev-...@lists.mozilla.org
> I checked the source code of the page. There is a meta tag in the <head>
> area for available languages that does not contain "eo".
That's because there is no Esperanto translation.
> area for available languages that does not contain "eo".
My message was about the wording and the possibility to automatically
choose another language from the list of Accepted-Languages (the HTTP
header that comes from Preferences->Content-Languages) or let the user
choose from the list of available languages.
Eduardo.
Michael Wolf
Jul 24, 2017, 4:33:51 PM7/24/17
to
Eduardo Trápani schrieb:
Michael Wolf
Jul 24, 2017, 4:34:32 PM7/24/17
to
Eduardo Trápani schrieb:
Oh sorry, I misunderstood you.
Michael
Michael
Francesco Lodolo [:flod]
Jul 25, 2017, 5:24:15 AM7/25/17
to Mozilla l10n Mailing List
(adding back dev-l10n)
> The language detection seems to be using the Accept-Language *and*
> general.useragent.locale. If you add 'eo' as the main language in
> Preferences->Content->Languages you should be able to get the warning.
It doesn't use general.useragent.locale.
I set intl.accept_languages to just "eo", website correctly shows up in
English as a fall back. But, since you have requested 'eo' as your
preferred language, it still warns you about Esperanto not being
available for this add-on. It doesn't really know about your browser UI,
in fact I see that warning and my browser is in Italian (asking for eo
content).
> About the second point. If my Accept-Language is "eo,it,fr" I expect to
> be served content in one of those languages, in that order (or en-US if
> none is available). But I get the warning, the detection doesn't go past
> the first language.
True. I'm not completely sure if that's a bug though: you indicated that
your preferred language is Esperanto, does it make sense to warn you
that the add-on won't be available in Esperanto?
> That would be good. But it's not about switching, it's just about
> choosing one. I know there can be a language mismatch between the
> browser and Snooze Tabs, because I have it in en-US in my eo browser,
> so, there is probable a way to let me decide which language I want, at
> enable time.
Experiments are add-ons, so you're asking to have smarter fallback, or
language selection for add-ons. While I agree, that's a problem much
bigger than Test Pilot :-\
Francesco
> The language detection seems to be using the Accept-Language *and*
> general.useragent.locale. If you add 'eo' as the main language in
> Preferences->Content->Languages you should be able to get the warning.
It doesn't use general.useragent.locale.
I set intl.accept_languages to just "eo", website correctly shows up in
English as a fall back. But, since you have requested 'eo' as your
preferred language, it still warns you about Esperanto not being
available for this add-on. It doesn't really know about your browser UI,
in fact I see that warning and my browser is in Italian (asking for eo
content).
> About the second point. If my Accept-Language is "eo,it,fr" I expect to
> be served content in one of those languages, in that order (or en-US if
> none is available). But I get the warning, the detection doesn't go past
> the first language.
True. I'm not completely sure if that's a bug though: you indicated that
your preferred language is Esperanto, does it make sense to warn you
that the add-on won't be available in Esperanto?
> That would be good. But it's not about switching, it's just about
> choosing one. I know there can be a language mismatch between the
> browser and Snooze Tabs, because I have it in en-US in my eo browser,
> so, there is probable a way to let me decide which language I want, at
> enable time.
Experiments are add-ons, so you're asking to have smarter fallback, or
language selection for add-ons. While I agree, that's a problem much
bigger than Test Pilot :-\
Francesco
Eduardo Trápani
Jul 25, 2017, 10:41:25 AM7/25/17
to dev-l10n
> available for this add-on. It doesn't really know about your browser UI,
> in fact I see that warning and my browser is in Italian (asking for eo
> content).
You are right, that's what's happening.
> in fact I see that warning and my browser is in Italian (asking for eo
> content).
> True. I'm not completely sure if that's a bug though: you indicated that
> your preferred language is Esperanto, does it make sense to warn you
> that the add-on won't be available in Esperanto?
the list of languages.
As I see it, the list of accepted languages is the list of preferred
languages. I have to list them in order of preference, but I'm stating I
can work with all them. So, I don't want to be warned if you fall back
to one of those. I don't need a warning to fall back from eu->fr, or
gn->es if those languages were explicitly set in that order in my browser.
If you have to fall back to en-US (_and_ en-US wasn't a preferred
language), then yes, a warning makes sense. But while falling back in my
own private list of working languages, well, then I simply wouldn't do
it. That's why I take the time to list the languages I can work with.
> Experiments are add-ons, so you're asking to have smarter fallback, or
> language selection for add-ons. While I agree, that's a problem much
> bigger than Test Pilot :-\
(there was a thread about it just some days ago about it). There are
more multilingual than monolingual speakers in the world, but the
interfaces sometimes favor the monolingual ones, for example by forcing
you to traverse full pages before you can change the language to another
one that you might prefer, because of the topic for example.
But going back to Test Pilot, it is on the right track about the
fallback. I mean, if I list eo, eu, it, en-US, I get the page in Italian
(what I expected) with a big warning about eo not being available. The
addon language doesn't change after the first install though.
So, for me right it's really all about the (type of) warning and wether
it's needed at all in case of falling back through the users' own list
of languages.
Eduardo.
Francesco Lodolo [:flod]
Jul 25, 2017, 10:49:58 AM7/25/17
to dev-...@lists.mozilla.org
Il 25/07/17 16.41, Eduardo Trápani ha scritto:
Italian, and the add-on is not available in Italian (possibly with
better wording).
It might be indeed a different user case if you're falling back to a
different language, e.g. website in Italian and warning about eo. Maybe
it should just look at the locale currently used for the website: if
you're reading the website in Italian, it should be safe to assume
you're OK with the add-on in Italian as well.
I'll add it to the open issue. Not sure code knows that, but that's a
technical detail to figure out.
For reference: the website uses l20n, and has smart fallback logic.
Hopefully we'll get there for add-ons and Firefox UI too.
That also raises the interesting problem of differentiating UI language
selection (something that doesn't exist right now) from content language
selection (accept languages).
Francesco
> So, for me right it's really all about the (type of) warning and wether
> it's needed at all in case of falling back through the users' own list
> of languages.
I think it's fair to warn the user if you're watching the page in
> it's needed at all in case of falling back through the users' own list
> of languages.
Italian, and the add-on is not available in Italian (possibly with
better wording).
It might be indeed a different user case if you're falling back to a
different language, e.g. website in Italian and warning about eo. Maybe
it should just look at the locale currently used for the website: if
you're reading the website in Italian, it should be safe to assume
you're OK with the add-on in Italian as well.
I'll add it to the open issue. Not sure code knows that, but that's a
technical detail to figure out.
For reference: the website uses l20n, and has smart fallback logic.
Hopefully we'll get there for add-ons and Firefox UI too.
That also raises the interesting problem of differentiating UI language
selection (something that doesn't exist right now) from content language
selection (accept languages).
Francesco
0 new messages