Talk Python to Me is a weekly podcast hosted by Michael Kennedy. The show covers a wide array of Python topics as well as many related topics (e.g. MongoDB, AngularJS, DevOps).
The format is a casual 45 minute conversation with industry experts.
Have feedback for the show? Send it to [email protected]. We'd love to hear from you.
The show is owned and operated by PDX Web Properties, LLC. (a small company run by Michael Kennedy).
Пікірлер
huge thank you for the shout out to Powertools for AWS Lambda library <3
Quart user here! Migrated from Flask last year and up to now I haven't encountered any issues. Super stable and works very well. Haven't tried plenty of the extensions..but the ones I tried works well too: rate-limiter, auth.
With all this tracing thing it sounds like you reinventing OpenTelemetry tracing. We have it on in production for getting insights on the execution paths, timings and parameters. For any span you can add any meta (user name, id etc for instance), you can enable auto-instrumentation for requests, SQL database queries, external requests and much more. And all of that is free and with great support from CNCF. Grafana, Zipkin, Jaeger all work with it.
Log streaming needs high frame rates.
Great Show! I always enjoy the guests that come to these python podcasts. It’s really cool to hear their stories.
So much fluff 🙄🙄🙄🙄🙄
Michael - have you ever had Corey Schafer as a guest on the show? I did a quick search and couldn't find anything.
Hi Mike M. Corey is great. I had him on a little while ago on episode 169 of Talk python.
Thanks so much for having me on, Michael! Had a blast talking to you!
great talk, thanks!
This is just freaking great stuff 😃🙌
We all know the power button is the best way to close VIM. 🤣
yes @carlton I totally relate; whole page reload thing with htmx😂😂😂😂😂😂😂😂that was hilarious for me
34:43 "import numpy as pd" I hope nobody actually writes that 😆
This is a great talk, but I found Pandas for everyone to be a great book on this for the non-dev python users.
What design patterns is most convenient to work with using Flet? I wonder if it is flexible. I'm a noob and don't have the capacity to analyse if this framework is flexible or not.
Wow for the content, so much less views. Learnt alot,
Request for content; Exact workflow of deploying Flet (and/or) Flutter app on appwrite, with user authentication and database. Preferably with Windows user experience taken into consideration too.
Host is rely unknowing
I had to try it out :-). I creted a vid about my experiace trying *UV* within azure machine learning (... conda): kzread.info/dash/bejne/h6qH1beKZq-2m6g.html🤓
I would love an integration with using uv or any python dependency tool with a docker image. I find the process of setting up a python environment within a docker image to be quite cumbersome with each project requiring fine tuning to get all your packages happy to be running within a container.
Looking forward to what Charlie comes up with regarding testing. (pytest)
I would say that you can test locally the lambdas with a basic config of Serverless Framework and the offline plugin.
This works well if you are trying to make money / general purpose use. If you are trying to make something novel, that would require some more advanced concepts or create a new technology, AI just isn’t built for that, but that would be more on the bleeding edge of technology itself. Hopefully people don’t just fall for constantly using AI, because it will slow down progress in terms of new technologies.
Well, I'm trying to build rather simple products to solve rather simple problems / use cases, so in my situation it works perfectly fine. It would not be adapted to a more complex project, of course. But the (AI) technology will also evolve. At some point, I genuinely believe that we'll be able to manifest apps from a clear brief. Natural language is definitely going to be the new programming language: kzread.info/dash/bejne/hWR6s8uqhrTRhrA.html
👍
2:45 I dont know about everyone else. But I care about async mongo a lot. Thats like the only thing i care about. Although i didnt used to use an ORM in my fastapi + mongodb applications . But seeing that there exists a library focused on pydantic and async mongo the two things i care about the most. I am surely gonna use it.
Nice video! I will be waiting to try!
Doing excursions in the JavaScript world, I don't understand why none of the previous package managers like Poetry or Pipenv had script runners builtin. Specifying every tool your project needs in a package.json is just so awesome in JS projects and an absolute no-brainer to have in Python aswell.
Here’s my use case: data scientist writing code for repeated one-off analyses (that differ a little bit time to time but not much) that’s trying to write better and easier to work with code. Is hatch a good fit for this? Should I design my project code like a package? I think I get overwhelmed with all the different features but would like to use hatch bc it seems like I could incorporate things easily later on (like pytest). Is it worth the investment or should I just stick to a more simple structure?
Yes I think Hatch would be a great fit for your use case! There would be no need to turn it into a package but rather just create environments as you see fit and scripts to invoke your code. Your use case (if a single file script with dependencies) will become much easier in the next minor release when PEP 723 is implemented.
Super Awesome!
Also man i dont even know if you guys would reply to these comments, but anyway im a complete beginner just starting web development and i chose Django to do it. Its great so far but working with static files like images is a pain. Why is it that Django doesn't support working with static files in production? Going outside of Django to 3rd party services for using static files like images is confusing and really disappointing considering everything so far has been amazing for me. Any word on why this is? is it a standard across all web frameworks or just the way Django does things and if its possible that future versions of Django will add this ability in a future version.
The reason as far as I understand is that Django is not the fastest at serving static files. I've had a few applications in production for years now where on some im serving the static files with nginx and one a few with whitenoise. In my opinion nginx feels faster but I've never had any issues with whitenoise
You guys know when we can expect the Django developer survey 2023 results to be posted?
I like ORMs but I like writing my own SQL and then using something to deserialize the results. I created my own .net ORM called EasyORM to do just that. I want my orm to create the connection and let me send it the sql and tell what it type of data to return. I just create models to represent what the results look like. Probably not the best way but works for me.
Such creativity! I love seeing art + science, really cool!
"I would rather have it shaky than impossible". I get what you mean, but hard disagree. I would much rather feel the pain upfront, than wonder why everything is broken afterwards.
Wanted to clarify something mentioned regarding PyPI repository information that is incorrect: It was claimed that packages can control the users under "Maintainers" on a project homepage by changing packaging metadata. That isn't correct, the list of accounts and avatars under "Maintainers" is controlled entirely by PyPI itself, so the accounts listed in that section are always correctly associated with the project. What can be controlled via package metadata is the "author" and "maintainer" email addresses.
Also regarding the "immutability" of PyPI: hashes are important because they exclude new artifacts being published to the same "release". So even if you have urllib3==2.2.0, someone could upload an artifact that is more specific (via platform tags, build numbers, many different options here, see PEP 440) that still matches urllib3==2.2.0 but would be favored over the old artifacts. If you were using hashes before the new artifact would be excluded from the set of artifacts that pip would evaluate as installation candidates. Without hashes pip would consider the new artifacts just the same as the old ones.
Great that microsoft is financing those two gentlemens effort in making python faster. Excellent Talk. Looking forward to 3.13.
I recently retired from the US Army, and although I have no clue what this talk was about, I would like to learn. The last time I used a programing language was basic, dos, hahahaha
Around 46:14 Steve says conda sits outside of Python. I don't understand. Conda is one of the most important and useful vehicles for end-users of Python in the scientific community. I don't understand how these discussions almost universally treat conda as some kind of weird anomaly, when it is used by so many people (ie Sara in the lab).
Are you guys going to make a course for Litestar? There are not much tutorials on this framework
Just listened to this episode. Excellent work on Rich! And I am loving objexplore - thanks for the tip!
Wonderful
I read that in “Broken” Matt Hardy’s voice 😂
Excellent overview of the topic
Love the content... 100%
Any GCP Cloud Docker connectivity from Pycharm IDE?
For retrieving exact AI/ML result set in template
Please provide me some resources about Jetbrains Pycharm Professional IDE Python Django methods and class that can be run partly for distributed Python package on my Windows 10 laptop
Sarah is veryyy excited! 😂😂
When is Sebastian going to do this for Dataframes e.g. pandas?
Very interesting talk and an outstanding guest! Hope to see more of Stas in the future!
Very nice, young and talented. I would like to see live stream with a person who was a career changer. Looks like 40+ people do not have much opportunities. Thanks