[Action] Automation to detect required website translations

[leonardpahlke]

The TAG ENV website now supports multiple languages. With this new feature, it's essential to ensure that not only the main English version is updated but also the translated versions. Currently, translators must monitor the website; if they notice any changes, they open a PR. This process is error-prone and stressful for those maintaining the translated content. Ideally, when a PR updates the English version of the website (any part of the website), a follow-up issue would be automatically created for each language, which our team can then address.

• This likely would require writing some part of custom code which we could host in the tag-env-tooling repository.
• The code could check over a GitHub action if artifacts of the website are touched and if this is done in a en folder.
• The GH action would then open an issue with details (labels translation required) and ideally ping (over github) a group of people maintaining updates.
• Optional: Add the opened issues to a new GH project board.

[SamYuan1990]

https://github.com/dorny/paths-filter
I hope by this github action we can have a change list, as a part of monitoring of the change.

is there any one knows how to create github issue automatically with a change list?

[kumarankit999]

I would also like to help here! @leonardpahlke

[sergiopsyalo]

I would also like to help too @leonardpahlke

[leonardpahlke]

Assigned you both! Thanks!

[guidemetothemoon]

+1 for getting some automation in place for this!

For the QA checks workflow we're using a changed-files GH Action that works really nicely on identifying changes in the specified files/directories so this or simple action can be relevant to use here as well. Implementation is in this workflow: https://github.com/cncf/tag-env-sustainability/blob/main/.github/workflows/checks.yml

There are multiple GH Actions to choose from that can be used to automatically create an issue based on some event. For example, this one: https://github.com/marketplace/actions/create-an-issue

Maybe we can even create groups with people that have been contributing to the different language translations so that those groups can be added on respective issues (f.ex. spanish translators group to an issue related to translating spanish content).

I'll be happy to contribute both on the implementation and review. Don't hesitate to reach out @kumarankit999 @sergiopsyalo 😊

[leonardpahlke]

@kumarankit999 could you join the next TAG ENV meeting (Feb 14) to briefly report on the status and perhaps we can discuss any questions you may have. https://calendar.google.com/calendar/embed?src=72e93a411f02e5664bb4485c04311b83dae6a62574e4ab882a1ccf8526aa9bf1%40group.calendar.google.com

[sergioarmgpl]

Sorry I commented with the wrong user, could you assign the user @sergioarmgpl instead @leonardpahlke

[leonardpahlke]

@kumarankit999 & @sergioarmgpl pls let us know if you need any assistance here! thanks :)

[sergioarmgpl]

I will use this actions as a reference https://github.com/marketplace/actions/create-an-issue I will need this part ${{ secrets.GITHUB_TOKEN }} I mean that secret to have super powers to create the issue cc: @leonardpahlke

[leonardpahlke]

I will use this actions as a reference https://github.com/marketplace/actions/create-an-issue

👍

AFAIK the GITHUB_TOKEN is a default environment variable which gets automatically set in GitHub Actions. Should exist by default.

[sergioarmgpl]

@Dianmz will be with me for some shadowing, just to introduce her to this kind of Issues :), Sure, I will use the GITHUB_TOKEN in that way :) @leonardpahlke

[sergioarmgpl]

Retaking this after completing the translation into Spanish

[SamYuan1990]

how is this issue on going?

[leonardpahlke]

how is this issue on going?

@sergioarmgpl ^ — do you need any support? Do we have a plan how to implement it?

[sergioarmgpl]

I will work on this, this week, sorry I was a little bit absent.

[thelooter]

Another way this could be approached would be using a tool like crowdin or Weblate to manage translations. You would essentially put in the translatable strings and then add languages it should be translated too. This would also make tracking the process of translations easier. It would also lower the bar to entry for translation, since you could much more easily translate small chunks instead of needing to translate a whole document at once

[leonardpahlke]

Thanks @thelooter!

@cjyabraham would be interesting to hear your thoughts on this.

[leonardpahlke]

cc @nate-double-u

[cjyabraham]

I think the place for human-crafted website translations is shifting as automated in-browser AI translations get better and better. If we're not already there, I expect we will soon be at the point where automated translations are about as good as those done by a human. Given that, I would look at supporting human translations as more of a short-term thing rather than something we'll need to do indefinitely. I'm interested, of course, in other people's views on this :)

We've also been exploring this topic wrt the Events site and, as of now, have opted not to build out a complex translation infrastructure.

Another temporary solution for this current issue is to create an issue every time an edit needs to be done to the site with checkboxes for each of the languages. Once the English language edit is done, the other language teams can be assigned to the issue until they have done their translation and checked their box. I'm not sure if this is practical given the cadence of English-language edits but it would save having to have GH actions running to automate things. I'm curious to know how the Glossary team manages this...

I don't like the idea of integrating with a 3rd-party service to have them manage our translations. Integrating sites like that is rarely simple and ties us to that closed-source solution.

[sergioarmgpl]

At least by my side just to detect different make the list of the files to update them send the notification, thats my plan, I have time to work this week, I have some vacations :) news soon.

[sergioarmgpl]

I saw into the glossary there is no automation for translation. So its made by humans. Also I think that human intervention promote people to contribute in some way.

[leonardpahlke]

Also I think that human intervention promote people to contribute in some way.

Yes, translations are a low-barrier type of contribution aimed at parts of the community that are not yet represented in the TAG. However, in general contributions should not be obsolete, otherwise they lose their value.

[thelooter]

I think the place for human-crafted website translations is shifting as automated in-browser AI translations get better and better. If we're not already there, I expect we will soon be at the point where automated translations are about as good as those done by a human. Given that, I would look at supporting human translations as more of a short-term thing rather than something we'll need to do indefinitely. I'm interested, of course, in other people's views on this :)

We've also been exploring this topic wrt the Events site and, as of now, have opted not to build out a complex translation infrastructure.

Another temporary solution for this current issue is to create an issue every time an edit needs to be done to the site with checkboxes for each of the languages. Once the English language edit is done, the other language teams can be assigned to the issue until they have done their translation and checked their box. I'm not sure if this is practical given the cadence of English-language edits but it would save having to have GH actions running to automate things. I'm curious to know how the Glossary team manages this...

I don't like the idea of integrating with a 3rd-party service to have them manage our translations. Integrating sites like that is rarely simple and ties us to that closed-source solution.

While I agree that AI gets better and better every day, its imo just not at the point yet, where it can be reliably used for translations. Many languages rely on certain phrases that can't just be translated one to one, and reasoning those phrases is still a big challenge for AI.

AI can maybe be used to create an initial translation that's refined by humans but I wouldn't just blindly let AI translate it.

I agree that binding ourselves to a closed source tool is not an ideal solution, that's why I suggested weblate. It's open source (https://github.com/WeblateOrg/weblate)

This also makes updating and tracking translations a lot easier since it's managed in smaller chunks and therefore easier to spot changes. As I stated before, it also allows for easier contribution, as one doesn't have to edit/translate a whole document but they can edit small snippets, e.g. while on the train.

[nate-double-u]

AI can maybe be used to create an initial translation that's refined by humans but I wouldn't just blindly let AI translate it.

This has been what I've heard from localization teams too. AI can be used for a first pass, but it doesn't quite reduce the human workload yet because of the amount of editing that is still required.

I'm curious to know how the Glossary team manages this...

@jihoon-seo & @seokho-son may have some insight on how the glossary localization teams manage this.

[sergioarmgpl]

This our advance, we already detected the differences between different the languages.
We found:

• Some inconsistencies in a file called landscape/SustainabilityUseCasesAndLandscape2023.ko.md, We would like to remove the ko, just to standardize the files a little bit.

The other thing is that I will create an Issue using a cron workflow in actions, notifying the active people working on Spanish and Chinese. Thats our idea.

Questions:

• What do you think about?
• What do you think about calling a script in the workflow or executing bash in the workflow?

@leonardpahlke @guidemetothemoon cc: @Dianmz

[Dianmz]

To follow up the previous comment, how would you like or preferred to visualize the changed files?

Personally I prefer to create an issue but we want to know your opinion.

cc: @sergioarmgpl @leonardpahlke @guidemetothemoon

[SamYuan1990]

This our advance, we already detected the differences between different the languages. We found:

• Some inconsistencies in a file called landscape/SustainabilityUseCasesAndLandscape2023.ko.md, We would like to remove the ko, just to standardize the files a little bit.

The other thing is that I will create an Issue using a cron workflow in actions, notifying the active people working on Spanish and Chinese. Thats our idea.

Questions:

• What do you think about?
• What do you think about calling a script in the workflow or executing bash in the workflow?

@leonardpahlke @guidemetothemoon cc: @Dianmz

for Chinese, please notify me :-)

[leonardpahlke]

Some inconsistencies in a file called landscape/SustainabilityUseCasesAndLandscape2023.ko.md, We would like to remove the ko, just to standardize the files a little bit.

We are currently in the process of rewriting the landscape document. This will take another month or two. We can remove the KO file and follow standards! (@seokho-son @sysnet4admin — we started to translate the TAG env website in different languages, like Spanish see PR, we could do the same for Korean)

The other thing is that I will create an Issue using a cron workflow in actions, notifying the active people working on Spanish and Chinese. Thats our idea.

Sounds good! We could create a GitHub team tag-env-translators-spanish, tag-env-translators-korean, tag-env-translators-german, tag-env-translators-chinese, … and tag them in the issue. This can be defined in the CNCF repository “people” that has a manifest file https://github.com/cncf/people/blob/c7f8625ebd386574959bcb807de1b48ded6f4da2/config.yaml#L1160 (example team: tag-env-chairs)

[leonardpahlke]

To follow up the previous comment, how would you like or preferred to visualize the changed files?

Personally I prefer to create an issue but we want to know your opinion.

I think issues are the best way for sure. One issue per language based on PR. So if we open a PR which makes changes to multiple files we just mirror that with one issue per language (not per file).

[guidemetothemoon]

Yes, I also vote for issues.
Suggestion regarding GitHub teams for this also makes a lot of sense.

[sergioarmgpl]

Thank you, retaken this issue this week, yesterday was my birthday and I left my job 🙃 cc: @leonardpahlke @guidemetothemoon

[sysnet4admin]

@leonardpahlke

We are currently in the process of rewriting the landscape document. This will take another month or two. We can remove the KO file and follow standards! (@seokho-son @sysnet4admin — we started to translate the TAG env website in different languages, like Spanish see #340, we could do the same for Korean)

I will take a look after finishing my duty work 😭

[sergioarmgpl]

Thank you, we have deadline with @Dianmz to finish this the next week. I will create the issues as suggested. cc: @leonardpahlke @guidemetothemoon

[guidemetothemoon]

Thanks for keeping us updated @sergioarmgpl and happy belated birthday!🥳 Don't hesitate to reach out if you need any support.

[Dianmz]

Take a look at our progress 👀

cc: @sergioarmgpl @leonardpahlke @guidemetothemoon

[Dianmz]

We are going to keep on working on a general workflow for any language to avoid reusing code

cc: @leonardpahlke @guidemetothemoon

[Dianmz]

We found some trouble with the workflow but we are basing on the Kubernete's glossary

cc: @leonardpahlke @guidemetothemoon @sergioarmgpl

[Dianmz]

We are currently testing the final configuration of the workflow!

very close to reach the goal 🎯

cc: @leonardpahlke @guidemetothemoon @sergioarmgpl

[guidemetothemoon]

Very exciting! Thanks for all your time and effort, and for sharing updates with the TAG @Dianmz @sergioarmgpl 💚

[sysnet4admin]

Some inconsistencies in a file called landscape/SustainabilityUseCasesAndLandscape2023.ko.md, We would like to remove the ko, just to standardize the files a little bit.

We are currently in the process of rewriting the landscape document. This will take another month or two. We can remove the KO file and follow standards! (@seokho-son @sysnet4admin — we started to translate the TAG env website in different languages, like Spanish see PR, we could do the same for Korean)

The other thing is that I will create an Issue using a cron workflow in actions, notifying the active people working on Spanish and Chinese. Thats our idea.

Sounds good! We could create a GitHub team tag-env-translators-spanish, tag-env-translators-korean, tag-env-translators-german, tag-env-translators-chinese, … and tag them in the issue. This can be defined in the CNCF repository “people” that has a manifest file https://github.com/cncf/people/blob/c7f8625ebd386574959bcb807de1b48ded6f4da2/config.yaml#L1160 (example team: tag-env-chairs)

Hi @leonardpahlke
I commit and push the content of ko.
There is partly translating contents. so maybe it is not poor quality ;)
So failed to quality check. It may consider to check before merging it.

Please let me know if I could contribute more for it.
Have a nice weekend!

[sergioarmgpl]

This is the PR for this issue: #407

[sergioarmgpl]

Hi Dear TAG Env Team, this is the final the work that we are doing with @Dianmz.

This includes a workflow called Check outdated content, that generates new Issues for Languages for ES & ZH languages at the moment. Currently the issues are assigned to Diana and I, but should be modified to the new groups that you suggested or new ones.

The issues will look like this:

And the issue like this:

The workflow is in the .github/workflows/check-outdated-content.yml

[Dianmz]

@sergioarmgpl and I have just finished the first version of this issue!
We will create a second issue to optimize the languages with a reusable workflow.

cc: @leonardpahlke @guidemetothemoon

[sergioarmgpl]

Also we have to update the assignees for the new teams to create in Github such as tag-env-translators-spanish, tag-env-translators-korean, tag-env-translators-german, tag-env-translators-chinese. This is pending cc: @leonardpahlke @guidemetothemoon

[claire-fletcher]

Hey! What is the status on this work? It would be great to get an update in the WG Comms meeting.

[sergioarmgpl]

Just getting back this Friday to my country I want to finish this, I will check for the errors

…

On Wed, Sep 11, 2024, 11:38 Claire ***@***.***> wrote: Hey! What is the status on this work? It would be great to get an update in the WG Comms meeting. — Reply to this email directly, view it on GitHub <#325 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AGJSZZU4R2CYE7IF7ZIIYQTZWAFPZAVCNFSM6AAAAABCRJVFIWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGNBTGE2DKMBRGE> . You are receiving this because you were mentioned.Message ID: ***@***.***>