This task will mostly be for @Trizek-WMF. After the copy for the new survey is written in T206375 (including questions, headers, buttons, and other text), we will need to translate it into Czech and Korean, and make sure that Czech and Korean respondents share an understanding with our team over what the questions and responses mean.
We will likely want to test these with multiple speakers of those languages.
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | MMiller_WMF | T206365 [EPIC] Growth: Welcome survey | |||
| Resolved | RHo | T206372 Personalized first day: design survey experience | |||
| Resolved | Trizek-WMF | T206376 Personalized first day: translate and test copy |
Stalled until the source messages are ready.
However, @MMiller_WMF I would advise to have those messages on Translatewiki as a separate group, so that we can potentially work with more communities.
@Trizek-WMF -- could you please tell us some more about whether we should expect to use Translatewiki for this, or to translate locally? @SBisson, @RHo, @Catrope, and I were wondering what to plan for.
Use translatewiki.net
Have local messages
Really? I thought there was no special deploy window for localized strings. IIRC i18n cache rebuild is done only on train and deployers tend to avoid it in SWATs, because it (IIRC) require full scap, which takes up to 45 minutes, while SWAT is one hour and should deploy 6 patches.
Have local messages
- Pros:
- we can try different messages for each wiki (translatewiki messages are translations of a source) and update them whenever we want (there is a delay from translatewiki). However, that works as well if we use translatewiki, while it is possible to override locally any message. (That a weak pro.)
- Cons:
- if the tool remains in production for whatever reason, a transfer to translatewiki will have to be made again (no idea on how complicated that is)
If you delete message from the en.json file, it won't be deleted automatically from translatewiki.net. If you think you need it again, you will not do the manual deletion part.
@MMiller_WMF, is the deadline for that task still relevant? IIRC, there is no text already available to translate.
Brief update: we will be loading the text to Translatewiki this week, hopefully within a couple days.
The Welcome survey messages are now in TWN under the ext-growthexperiments-welcomesurvey message group.
Hi Stephane, why is welcomesurvey-sidebar-editing-link3-text and others in https://github.com/wikimedia/mediawiki-extensions-GrowthExperiments/blob/master/i18n/welcomesurvey/en.json empty? It caused empty messages on Translatewiki.net created. If we do not want to use those messages anymore, we should probably delete them from en.json (and qqq.json, if present) and inform TWN to delete those messages from their system (IIRC messages won't get deleted otherwise).
Hi Martin, the 4 messages related to editing link 3 and link 4 are left blank intentionally. We have messages for 4 links in the "Getting started with editing" side panel but we're only using 2 of them at the moment. They are defined so we can quickly have between 0 and 4 links on different wikis without changing the code. Each link has a message that ends with -text and one that ends with -title.
You can submit blank (like de has done) or ignore them (like ko has done) and it's fine.
@revi and @Urbanecm -- we now have the survey available in the Beta Cluster. It would be great if you could try it out. We're interested in your opinions on all the parts, but in particular, we want you to make sure the translations are all rendering correctly, and nothing is cut off, and that they make sense in the context of the survey. Here are the links for your two languages:
Would you be able to do this by Tuesday, 2018-11-13?
Confirmed.
No, because the links you gave us doesn't work. They redirects to the main page. How can I check it if I cannot view the survey?
@Urbanecm -- sorry about that. I think what happened is that your account is in the control group (the group that does not get the survey, for the purposes of our experiment. To make sure you get the survey, you can add &welcome-survey-experimental-group=exp1_group1 to the URL.
So like this:
Does that work?
Beta Labs is not happy with me - my first session failed and I needed to create another account, which prompted me to log in again immediately after I created an account and clicking 'done' prompted me to log in, again...
But whatever, I found one problem with translation (fixed now).
And I'm wondering - if account creator (or admin) creates an account for someone else, do they see this page when users log in for the first time? I am periodically creating 5 to 10 accounts every two weeks or so as part of outreach events, and I'm wondering if they will see this survey.
Everybody, it is not limited to admins. Admins can only bypass six account limit. Just enter Special:CreateAccount to the search box on any wiki while you're logged in.
When you create an account that way, IIRC the user has to log-in. If we can detect if that's the first connexion for the account, we can display the survey. I've created a sub-ticket: T209583: Accounts created using Special:CreateAccount should trigger the survey at the first connexion
Accounts created on behalf of someone else are excluded from the survey. I don't know if they fit the population we are currently looking at. I think maybe "retention among users who start by attending a workshop" is an interesting but different research question.
From @nettrom_WMF