Generate a translation priority list

[jacoblubecki]

Suggestion(s)

1. Use any analytics that might exist for page traffic to figure out which wpilib pages are the most used and create a list ordered from most to least visited and use that as a reference order for translating pages. This would probably need to be regenerated at the beginning/end of each build season.
2. If that data is difficult to use or non-existent, a fallback might be doing some sort of site-mapping exercise and assigning some heuristic based on depth of resource (e.g. how many links you'd have to click to get there from a top-level link to get to the page) and/or based on prevalence (e.g. how many pages link back to a page).

This would help translation teams make the most impact early on and avoid duplicated effort on new/high-churn pages.

---

Context

I have been leading the translation for the Japanese frc-docs for the past several months. A common thought I've had is "it's difficult to figure out what to translate next". When there's not a clear answer, I usually ask around until we settle on something, but this isn't a great use of time and doesn't guarantee we are working on a page that is actually that useful.

My manual process has basically boiled down to:

1. Translate explicit zero-to-robot resources
2. Translate resources that are referenced in the indices for zero-to-robot
3. Translate random pages that are referenced in zero-to-robot pages
4. Translate pages that feel important
5. ???

Not to say this is a terrible approach, but we are getting to a point where we are spending relatively a bit more time pondering on what to work on vs. actually working on things.

I also think it's helpful for delegating and motivating. I would always have the next item ready to go when I have a translator who has time to help, and translators always know they are working on the most important thing that needs to be worked.

[Daltz333]

You pretty much have it. In the past, we've delegated the actual priority list to the language coordinators. This is for a couple of reasons.

• minimizes development bandwidth on us
• gives the translation teams for various languages maximum flexibility
• goals may be different per language

From a growth perspective, it's typically important that the "upper" level content of the docs

• ztr
• vscode basics
• tools/dashboards
• driverstation

get translated first.

[jacoblubecki]

I think just having that basic list in the "contributing to frc-docs" page would be helpful. We spent a bit of time translating Romi documentation, for example, because it was listed near the top of the list using the Transifex priority sort.

That said, a major motivation for this ask the number of things like:

See <xyz> for more information

It feels a bit like leaving a page unfinished if it references supplementary documentation or similar. A recent one we dug up was the can wiring basics page which explains how to do CAN termination and is referenced in the robot wiring guide -- while we covered a lot of the wiring guide, any team that was redirected would end up having to figure it out themselves.

Another we missed was the glossary which is referenced in a bunch of pages via the glossary tooltip things.

[Daltz333]

Hm, yeah, the translation priority list isn't that useful I feel like. When I mentioned "top-down", I more mean top-down on the actual https://docs.wpilib.org/ site itself.

Mentioning this general guideline is probably useful to add to the translation page. I agree.

[jacoblubecki]

I think that's an okay proxy, but it would be helpful to mention in the contribution guidelines. Something like this would have been super helpful for my translation team:

We recommend starting with translating the "zero-to-robot" series. Once that is finished, it may be useful to work your way down the side bar from top to bottom, as there is a good correlation between a page's usefulness and how high it is on the sidebar (items at the top of the sidebar being the most useful).

While Transifex does include a priority sort, it is not very accurate and we would recommend against using that when starting out. That said, translation teams are free to translate in whatever order works best for them.

Otherwise, the Transifex "priority" sort is likely to lead other new translations to wasting time like it did for ours.