Recommended repo for hosting new Docling converter? #8614
-
|
Hi Haystack team, we would like to add 🦆 Docling document conversion capabilities to Haystack. We understand the abstraction for that would be a Converter, however we notice that different converters have been placed in different repos, e.g.:
Is there some specific recommendation where / how to host our new converter — or would it be up to us to pick an option? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment 1 reply
-
|
Hello @vagenas Thanks for reaching out! We would recommend that you implement a Docling converter in your separate GitHub repository so that you remain the owner of the code and have full control when it comes to releasing new versions of your integration. Feel free to reach out to us if you would like someone from our open source engineering team to provide guidance once you start implementing the integration or to have a look at the code. Discord works well for that or we can arrange a call too. Let me provide some more context regarding the different repos. haystack-core-integrations contains Haystack integrations that our open source engineering team is actively maintaining. A very limited number of converters/integrations is actually part of the core Haystack repo. They don't need any extra dependencies installed and work out of the box, which is why we kept them in Haystack core so far and didn't move them to haystack-core-integrations Looking forward to your Docling converter! |
Beta Was this translation helpful? Give feedback.
-
|
Many thanks for the pointers @julian-risch! PR just submitted: deepset-ai/haystack-integrations#292 |
Beta Was this translation helpful? Give feedback.
Hello @vagenas Thanks for reaching out! We would recommend that you implement a Docling converter in your separate GitHub repository so that you remain the owner of the code and have full control when it comes to releasing new versions of your integration. Feel free to reach out to us if you would like someone from our open source engineering team to provide guidance once you start implementing the integration or to have a look at the code. Discord works well for that or we can arrange a call too.
Once the implementation is done, we'd be happy to list your integration on our overview page: https://haystack.deepset.ai/integrations For that, you just need to open a PR in the following repos…