Sofia
Account created on 11 February 2008, over 17 years ago
#

Merge Requests

More

Recent comments

🇧🇬Bulgaria valthebald Sofia

Option to set the default state for the new translation was already added

🇧🇬Bulgaria valthebald Sofia

@anmolgoyal74: that's a good proposal! Do you want to create a merge request, or someone else can?

🇧🇬Bulgaria valthebald Sofia

I've reworked original idea, and added the logic to hide "AI translations" column when user doesn't have permissions or all languages are already translated

🇧🇬Bulgaria valthebald Sofia

I have added "beta blocker" tag to all the mentioned issues, feel free to add tag/untag/add more issues

🇧🇬Bulgaria valthebald Sofia

Merged into 1.2.x, thanks everyone!

🇧🇬Bulgaria valthebald Sofia

@yuri.p: common flow is to target the current default branch (1.2.x), and then, if needed, backport to previous one (1.1.x)
There's almost no chance new commits will be added to 1.0.x

🇧🇬Bulgaria valthebald Sofia

@IliaNoz: any reason you changed the target version from 1.2.x to 1.1.x?

🇧🇬Bulgaria valthebald Sofia

@IliaNoz: any reason you changed the target version from 1.2.x to 1.1.x?

🇧🇬Bulgaria valthebald Sofia

Lowering down the version, since we already have enough issues in the work, and it would be good for people to understand which issues go where

🇧🇬Bulgaria valthebald Sofia

valthebald created an issue.

🇧🇬Bulgaria valthebald Sofia

This is fixed in Content moderation options in entity translations Active . If you want this feature to be 1.1.x, please provide a backport

🇧🇬Bulgaria valthebald Sofia

@dmundra: thanks for the update! Checking soon

🇧🇬Bulgaria valthebald Sofia

I've made several commits to the repo with the following results:

  • Added cache tags to the calendar view. That fixed stale cache when issues are added/updated
  • Added d.o. userid field to contributor content type. This reduced the number of calls to the API
  • Added static caching of user and node data when processing a project
  • Added drush commands to process single module and all modules
  • Process single module configuration using Drupal queues. First thing that is done by drush command is populate the queue, so even if the drush command fails for any reason, the queue can be processed later
  • Handle "429 Too many requests" responses from API (up to 3 retries with 30 seconds delay before finally giving up)
  • Added contributor now shows in the dashboard

Regarding timeouts, the proper way to overcome this would be to run the imports in scheduled cron jobs, but unfortunately, DF does not support that yet.

🇧🇬Bulgaria valthebald Sofia

@anjaliprasannan: I could reproduce the issue locally with the latest 1.2.x, and your patch fixes it.

Merged to 1.2.x, thank you!

🇧🇬Bulgaria valthebald Sofia

Assigning to myself to check how issue assigning works on AI dashboard

🇧🇬Bulgaria valthebald Sofia

I get this error: TypeError: Gemini\Data\GenerationConfig::__construct(): Argument #6 ($topK) must be of type ?int, float given, called in /var/www/html/web/modules/custom/gemini_provider/src/Plugin/AiProvider/GeminiProvider.php on line 260 in Gemini\Data\GenerationConfig->__construct() (line 39 of /var/www/html/vendor/google-gemini-php/client/src/Data/GenerationConfig.php).

(probably related to the signature change in the library between the versions).

Steps to reproduce:

1. Enable AI Explorer module
2. Go to AI Explorer > Chat generation
3. Select any tool

Here's my configuration:

🇧🇬Bulgaria valthebald Sofia

Going to check this today and report the results

🇧🇬Bulgaria valthebald Sofia

Merged with 1.2.x, thank you @Aporie!

🇧🇬Bulgaria valthebald Sofia

Added some comments on PR (first one is a deal breaker)

🇧🇬Bulgaria valthebald Sofia

Merged with 1.2.x, thank you @anjaliprasannan

🇧🇬Bulgaria valthebald Sofia

@ralkeon: can you please create a merge request? and what is your overall configuration (i.e. steps to reproduce)
Thank you!

🇧🇬Bulgaria valthebald Sofia

We can use Add text extractor plugins for image and link field types Active as a starting point. With the new base class for field extractors, adding a new one for address field should be quick.
Also, bumping version

🇧🇬Bulgaria valthebald Sofia

I don't think we need to wait until 2.0 to have this feature, downgrading target version

🇧🇬Bulgaria valthebald Sofia

I like the idea, only small changes in MR please

🇧🇬Bulgaria valthebald Sofia

@anjaliprasannan: I don't think so. LB itself has own abstraction layer and does not depend on block_content.
The issue is ai_translate-specific, and should be fixed in LB extraction plugin

🇧🇬Bulgaria valthebald Sofia

@Aporie nice touch!
Can you please review the messages in https://git.drupalcode.org/issue/ai-3536092/-/jobs/5926826, and also ensure that the code applies to 1.2.x? Since this is schema changing, we probably need an update hook to change existing configurations, especially is this will be backported to 1.1.x

🇧🇬Bulgaria valthebald Sofia

This issue has bothered me "in the background" for some time, only to realize that translations actually can be set to unpublished even when the source entity is published, so this is handled by content moderation module somehow.

So I debugged what happens in content moderation module and added the same code to AiTranslateController.

Please review

🇧🇬Bulgaria valthebald Sofia

@marcus created 📌 Move TextExtractor service to AI core Active and assigned to myself. Will try to submit a proposal by Monday EOD

🇧🇬Bulgaria valthebald Sofia

@svendecabooter I haven't experienced this, will check

🇧🇬Bulgaria valthebald Sofia

@svendecabooter: done! Thanks for pointing this out

Production build 0.71.5 2024