Closed Bug 1446891 Opened 2 years ago Closed 2 years ago

[crh] Search engine setup for Firefox for Crimean Tatar

Categories

(Mozilla Localizations :: crh / Crimean Tatar, enhancement)

enhancement
Not set

Tracking

(firefox61 fixed)

RESOLVED FIXED
mozilla61
Tracking Status
firefox61 --- fixed

People

(Reporter: flod, Assigned: flod)

References

()

Details

(Keywords: productization)

Attachments

(1 file)

We want to ship Firefox with a good list of search engines for Crimean Tatar. 

We'll do this in three phases: First, the localization team and flod (Francesco Lodolo) will look at the market for the language and come up with a good set of search providers. Reşat, we'll need your input on this. The guidelines for making recommendations are found on <https://wiki.mozilla.org/L10n:Firefox/Productization>. 

We currently use this set for en-US, to which we can fall back in the case of lacking localized alternatives:
Google, Yahoo!, Bing, Amazon, DuckDuckGo, Twitter, Wikipedia, eBay

Adding brand new search engines, that not yet ship in any other language, needs approval from Business Development and requires more time to get authorization and assets from the provider. To avoid blocking the initial set up, this type of requests should be split into follow up bugs. Reşat, we'll likely need your help to at least get the owner's contact information.

Once we have a specification on what the search engines for Crimean Tatar should look like, flod and the localization team will work on an implementation on that. flod will create and attach patches for the desired changes and get them reviewed. After a successful review, flod will land them. Reşat, if you're interested, you could create the patches, too. You'd ask flod for review on those before landing.

Please don't commit any modifications in browser/searchplugins to the hg repository without a patch reviewed by flod, Stas, Axel or someone else appointed by flod. The data here is sensitive to users and our search partners, thus requiring extra care here.

The Crimean Tatar team and flod will be driving this bug to completion.
For reference, these are the settings for Turkish
https://transvision.mozfr.org/productization/?locale=tr&product=browser

I see there is a Wikipedia version, so we should use that instead of Turkish, since it's what we usually do
https://crh.wikipedia.org/wiki/Ba%C5%9F_Saife

Should we replicate Turkish, and use the localized version of Wikipedia? The only reason for not using the localized version would be that it doesn't have enough content.
One more note: I need to check with our BD about the default search engine.
(In reply to Francesco Lodolo [:flod] from comment #1)
> Should we replicate Turkish, and use the localized version of Wikipedia? The
> only reason for not using the localized version would be that it doesn't
> have enough content.

Thoughts on this?
Component: Other → crh / Crimean Tatar
Flags: needinfo?(haqer)
@Reşat
Note that we can't start Nigtly builds without setting up search engines.
(In reply to Francesco Lodolo [:flod] from comment #1)
> For reference, these are the settings for Turkish
> https://transvision.mozfr.org/productization/?locale=tr&product=browser
I've taken a look at az, kk, tr, & uz.
> I see there is a Wikipedia version, so we should use that instead of
> Turkish, since it's what we usually do
> https://crh.wikipedia.org/wiki/Ba%C5%9F_Saife
I agree: & this is the only one i've been aware of from this PoV.

> Should we replicate Turkish, and use the localized version of Wikipedia? The
> only reason for not using the localized version would be that it doesn't
> have enough content.
In general, i think this is the best that can be done at this time (to the best of my knowledge), but i would suggest 1 more modification to inheritance from Turkish:
In Search order, I would suggest that the default not be Yandex, but be Google, for instance. A quick note on the future: once its investor recovers the costs of initial investment, most likely there will be a general (not just an mobile-app-based) Yaani search engine (https://yaani.com); at that time, it would probably be the best match for the default in search order.

P.S. Along with another fellow I tried to contact Google some years ago about localization. At that time, they had some voluntary l10n contribution options, & the thought was to set that up for crh as well. But no response from Google has been received. Something like this could be considered again in the long run.
P.P.S. There are fairly big online retailers in Turkey, but they aren't set up for Firefox at present (as part of the e-Commerce Search category). If they are set up in the future, inheriting them too would be fine i think (because, they might come in handy for users there).
Flags: needinfo?(haqer)
(In reply to Reşat SABIQ (Reshat) from comment #5)
> In general, i think this is the best that can be done at this time (to the
> best of my knowledge), but i would suggest 1 more modification to
> inheritance from Turkish:
> In Search order, I would suggest that the default not be Yandex, but be
> Google, for instance. A quick note on the future: once its investor recovers
> the costs of initial investment, most likely there will be a general (not
> just an mobile-app-based) Yaani search engine (https://yaani.com); at that
> time, it would probably be the best match for the default in search order.

I need to verify with our BD department if it's OK to not have Yandex as default.

As for additional search engines, they can be added (not as default), but we'll need to go through a difference process (it requires written authorization). Default search engines depend on commercial agreements, and right now we only have Google, Yandex and Baidu, depending on the region/locale.

@Joanne
Can you confirm if it's OK to use Google as default for this locale (Crimean Tatar). It's spoken across different countries in the area, and it's likely not of interest for Yandex.
https://en.wikipedia.org/wiki/Crimean_Tatars
Flags: needinfo?(jnagel)
Flod,

I need to review the contacts with Legal, I will follow-up as quickly as possible.  Thanks, Joanne
Flags: needinfo?(jnagel)
Hi, following up on your question.  Do you know what % of users of Crimean Tatar would be downloading in Turkey?  Based on the Wikipedia entry, it seems significant.  If that is the case, then we want to be sure we are adhering to our agreement with Yandex in default.
Flags: needinfo?(haqer)
(In reply to Joanne Nagel from comment #8)
> Hi, following up on your question.  Do you know what % of users of Crimean
> Tatar would be downloading in Turkey?  Based on the Wikipedia entry, it
> seems significant.  If that is the case, then we want to be sure we are
> adhering to our agreement with Yandex in default.
In general, this is a difficult question. I can only you a range, & a realistic prediction for the first few years.
1. Range:
1.1. Lower bound: single (maybe double) digits (based on usage of the unofficial extension). 
1.2. Higher bound: there are estimates that there about 6 million people of Crimean Tatar origin in Turkey. That would mean 
6÷80.8=7.4%
of users in Turkey. However, Crimean Turkish is not taught in schools, so 7.4% will never in practice be achieved IMHO. Due to this factor, & limited amount of publications & media in Crimean Tatar (in general in the world), I would say the higher bound can't be expected to be more than 
3.7%
(& even that is most likely not achievable (at present)).
2. Prediction (for at least the first few years).
IMHO, during the first 12 months, it might be single digits. During the 2nd year, it might become double digits. During the 3rd year, triple digits may be achieved (but i wouldn't be able to promise that).

P.S. Bottom line: i do not believe this number of users is significant enough from the viewpoint of any global/regional commercial agreements between big commercial entities. And if at any point in time, the number of users reaches a threshold that can be considered significant from that PoV, then the default can be quickly changed, & effectuated within the max of 6 weeks (rapid release calendar).
P.P.S. I can mention another example that might be informative:
a person from Russia considered over 20000 people of Crimean Tatar origin in Romania so insignificant, that he hid the keyboard layout added for the Linux env. & meant to simplify typing the texts (with letters like ĭ & í reflected for instance) in Dobruca dialect; this keyboard layout can since then be activated by some hack (basically typing some commands on the console (as opposed to selecting it like a normal layout from the UI))[1] along with other personal-type layouts (that was the state of affairs at that time: i haven't looked at it since). This is the kind of attitude i've personally become accustomed to expect from Russia...
[1] http://tilde-birlik.sourceforge.net/?p=Qirim/linux/klavye/Dobruca-Q&l=crh@haq
Flags: needinfo?(haqer)
Typo fix, i meant:
I can only give you a range, & a realistic prediction for the first few years.
It might also be informative to add the following. The crh localization now (the part that is complete) is definitely competitive qualitatively with tr localization. However, in general it would be tough for Crimean Turkish to be competitive with (Turkey) Turkish in Turkey, even though the langs are mutually comprehensible (once a person learns a few rules (which i summarized on 2 A5 pages in a couple of my little books)). Even if there was a Crimean Turkish TV channel, & a Crimean Turkish daily newspaper, Turkish language there would be tough to compete with. In particular, Turkish movies are highly addictive. So i guess, there'd also need to be dubbing going on. It would be a challenge even if one had a bunch of money available to invest in such endeavors: & in reality there isn't much investment in such endeavors at all.
For these reasons, etc. the higher bound mentioned earlier would be tough to achieve in real world.
There are some legal questions around search we are working to resolve. We will have an update next week.
Yandex will not be included, Google will be the default.
(In reply to Mike Kaply [:mkaply] from comment #13)
> Yandex will not be included, Google will be the default.

Thanks Mike, I'll prepare the patch and ask you for review.
Assignee: nobody → francesco.lodolo
Remove Yahoo and Bing from search order, since we don't plan to use them
https://hg.mozilla.org/l10n-central/crh/rev/ac619fd4932d2865b177153f52708c9c4c3e7af8
Comment on attachment 8968970 [details]
Bug 1446891 - [crh] Search engine setup for Firefox for Crimean Tatar

https://reviewboard.mozilla.org/r/237672/#review243430
Attachment #8968970 - Flags: review?(mozilla) → review+
Pushed by francesco.lodolo@mozillaitalia.org:
https://hg.mozilla.org/integration/autoland/rev/615a28d921ff
[crh] Search engine setup for Firefox for Crimean Tatar r=mkaply
https://hg.mozilla.org/mozilla-central/rev/615a28d921ff
Status: NEW → RESOLVED
Closed: 2 years ago
Resolution: --- → FIXED
Target Milestone: --- → mozilla61
You need to log in before you can comment on or make changes to this bug.