I really like what @mitchellh has been doing on ghostty PRs. Most of them get an “AI disclaimer” section in the description. Most of the time it’s something like “I architected this and AI wrote the tests”, but every once in a while you’ll see “AI wrote this and I reviewed it.” In those cases he’ll include a link to the whole session for others to see as well.
@spatie_be Command
There's some complexity in the command to handle efficient batching to Redis for large datasets. Other than that though it's not hard to wire up a simple command to kick off migrations for large swaths of data or specific models.
Distilling some helpful concepts from a large data migration project I worked on recently.
This was a Laravel project were we had to migrate the usual suspects of ecommerce to another platform (e.g. users, products, orders, etc)
@spatie_be Job
Designed to be idempotent so they could also be used for observer events (topic for another day). Key here is to use the transformation manager to grab the payload and send it on it's way.
Happy Festivus! For this years celebration I'd like to air some additional grievances. Somewhere during the rise of GraphQL, edge computing, and serverless tech we've lost the json:api spec which was a truly solid way of crafting APIs. I wish it was more common.
On this Festivus I’d like to air some grievances.
Please stop using Mongo as your primary and/or default data store. Just use Postgres and push data to document storage later on if you really need to.
Happy #Festivus!
The downsides here are CodePipeline, which may be a skill issue, and the complexity in the compute layer between application provisioning and application start up. Serverless Framework deployments are also slow here, but that's an application specific issue.
A solid server side framework along with an RDS instance and an asynchronous task queue (e.g. Redis + Celery) will get you a long way. Add in CodeDeploy and you have a great setup for 3-5 environments. S3 + CloudFront has been fantastic for static and single page applications.
On a funnier note, I need to convert between PIL and OpenCV often in this app depending on what transformations need to happen. OpenCV defaults to BGR while PIL defaults to RGBA and if you forget to convert, you end up with images that look like Yandu from Guardians of the Galaxy
I've been working on an image processing application for a number of months. I found the following article informative in explaining how images can be treated as numpy n-dimensional arrays.
https://t.co/ley38MQAlR
#pythonprogramming