Actions are not working

[Keyrxng]

All of the compute.yml runs are failing for this plugin which means that this plugin is not actually in use at all

Afaik we are using shivs DB and they are running a script to periodically update their DB with comments from the Ubiquity orgs. I don't know where that script is or what exactly it's fetching to replicate the DB which I need to be able to build ubiquity-os-marketplace/ubiquity-os-kernel-telegram#17

This should be resolved by the following:

• Fixing this plugin and posting to a Ubiquity DB
• Adding the script which backfills a user' DB with everything (keeping in mind, a partner will also run this when they install UbiquityOS to build their embeddings for the rest of the plugins that need em)
• Potentially storing a CSV that contributors can load into their instance so they don't need to pay for embedding thousands of comments

P.S: I have access to the db_dumps workflow that I wrote about a year ago, so I can luckily seed my DB with the prod stuff for qa and development which I'd have done in this scenario if it was available. My current dev DB was seeded with prod but is like 5 months stale now.

[Keyrxng]

@sshivaditya2019 I've messaged you on TG too, could you share the CSV or the script with me? Preferably the CSV but both would be nice. Thanks man

[shiv810]

@sshivaditya2019 I've messaged you on TG too, could you share the CSV or the script with me? Preferably the CSV but both would be nice. Thanks man

I sent you the CSV over Telegram. For the script, you can check it out here: https://github.com/sshivaditya2019/IssueScraper.

[Keyrxng]

@sshivaditya2019 I've messaged you on TG too, could you share the CSV or the script with me? Preferably the CSV but both would be nice. Thanks man

I sent you the CSV over Telegram. For the script, you can check it out here: https://github.com/sshivaditya2019/IssueScraper.

Thanks. I must have missed the task for building a scraper and doing things this way instead of using the plugin, sorry about that.

[0x4007]

@sshivaditya2019 I've messaged you on TG too, could you share the CSV or the script with me? Preferably the CSV but both would be nice. Thanks man

I sent you the CSV over Telegram. For the script, you can check it out here: https://github.com/sshivaditya2019/IssueScraper.

I noticed that you have some exclude keywords for the comments.

Best if you just exclude every bot comment.

We should only keep track of human comments and then eventually cross reference with their XP in the repo so that their comment has more weight

[Keyrxng]

Error creating comment > Successfully created comment

https://github.com/ubiquity-os-marketplace/text-vector-embeddings/blob/development/src/handlers/add-comments.ts#L16

The logic is throwing the error comment and then erroneously logging a success comment, this happens in more places than one and should be fixed.

---

ubiquity/work.ubq.fi#168 - I just created this task and this looks like the run for it https://github.com/ubiquity-os-marketplace/text-vector-embeddings/actions/runs/11987717282/job/33422202579

• it created an issue
• found no similar issues
• didn't suggest suitable contributors

I added a comment and the following run succeeded in adding that comment it appears.

There could be improved metadata logging to ID which tasks and comments the run is targeting.

[shiv810]

I added a comment and the following run succeeded in adding that comment it appears.

There could be improved metadata logging to ID which tasks and comments the run is targeting.

All of these issues should be fixed in the linked PR. Action Run

[Keyrxng]

I added a comment and the following run succeeded in adding that comment it appears.
There could be improved metadata logging to ID which tasks and comments the run is targeting.

All of these issues should be fixed in the linked PR. Action Run

I see ✓ Created Comment with id: IC_kwDOMw-7xM6S_V-H

Would be better if you logged the html_url of the issue/comment, it contains the owner/repo and gives a quick link to the target. You should fill the 2nd param of logger.method("string", {}) with useful debugging data url, title, body etc..

[shiv810]

Would be better if you logged the html_url of the issue/comment, it contains the owner/repo and gives a quick link to the target.

Don't you think that would be redundant? We could simply add it to the explorer and find the same information there. I think adding that would make the logging unnecessarily cluttered.

[Keyrxng]

Would be better if you logged the html_url of the issue/comment, it contains the owner/repo and gives a quick link to the target.

Don't you think that would be redundant? We could simply add it to the explorer and find the same information there. I think adding that would make the logging unnecessarily cluttered.

https://github.com/ubiquity-os-marketplace/daemon-disqualifier/actions/runs/11988400340/job/33423618955

If you look at other action plugins, we are extremely verbose and log as much useful and helpful data as possible, so no I think it's inline with what we normally do. It's free to log and actions log runs are only ever investigated for debugging purposes so the more the better.