Phildini @ Home @phildini@xoxo.zone

Finally a good way to run kubernetes: https://github.com/learnk8s/xlskubectl

Hello! Are you currently on #Patreon, and the primary thing you deliver is access to a community on #slack or #discord or #circle or #whatsapp?

If so, I'd love to talk to you, I'm working on something to help with that.

In case y'all are curious what it's like to work with @drewdil at @burbteam

#startuplife

🤣

It irks me a bit that @tailwindcss for #rails is apparently self-contained, but the #django package for tailwind requires npm?

Anyone have a django build that gets tailwind working without node?

A thing I learned about me recently is that I never crawled. I went from flat to using furniture to haul myself up and start walking.

So if it ever seems like “you have to learn to crawl before you walk” doesn’t seem to stick to me, well…

A thing I just said to @rosemaryforyou that's going to relate heavily to @braintapepod and the eventual next project:

I’m covering the “ok, but here’s how this relates to the Council Nicaea” angle, so you gotta pick up the “yeah, but it’s also homestuck and achewood” line

To be clear, @burbteam isn't going to be moving off #django any time soon. And the docs are _fabulous_ at getting you from Zero to One.

But we'll all benefit from more guidance one 1->2 and 2->10.

8/8

There are probably tools and packages that do this for me, I'm sure, so _really_ what I'm railing against here is education.

#django wants to be the framework for perfectionists with a deadline, but could be doing a way better job prepping devs for scale.

7/n

(Yes, those 100k inserts should be in background task and they are. I have related ire for celery, which will be a different thread)

6/n

The #django ORM knows about the DB connection, and if its in a transaction (ie, most requests), then it should be able to inspect and spit out a log that warns you about "hey this transaction has 100k inserts, are you aware of objects.bulk_create"?

5/n

The crappy orange-site answer to this would be "lol inspect your logs or hire a DBA/SRE"

What's amazing about that answer is how startup-unfriendly it is.

4/n

But @burbteam, we're ingesting hundreds of thousands of rows of data (soon to be millions), and almost every perf issue we've had is because we were ingesting those points as single INSERTs.

3/n

By this I mean: If you follow the #django tutorial (or the sqlalchemy tutorial, or the ActiveRecord tutorial), you're lead down the path of mostly operating on single objects / records and this is fine so long as you're dealing with data in the "< 1k rows" territory

2/n

A thing that bugs me about #django (and really all ORMs) that we keep running into @burbteam -- no graceful ramp between single-fetch and bulk processing.

1/n

One of the joys of hosting the site that hosts the searchable archive of Alameda Minutes is that I can figure out what the hell is going on. (Really anyone can and honestly more people should)

The issue on the table: ASPA's management of Franklin Pool

#alamtg 1/n

How I feel waking up late on the West Coast and logging on to the internet this morning

I don't exactly know how to tell friends who aren't parents that when they say things like "You signed up for this!" in response to any complaint, what they're saying to me is:

"Please don't actually share anything about your life".

Hey @simonw you've seen this, right?

https://replicate.com/blog/run-stable-diffusion-on-m1-mac

Run Stable Diffusion on a mac without CUDA 😱

The @flydotio project is the J.E.E.P. of Heroku and perfect for it.

As always, huge thanks to @simonw and @datasetteproj for making this possible, and all the code for how I did it is on Github: https://github.com/phildini/berkeley-data

Thanks! Share and enjoy. ✨