quentin.pradet.me
I like maintaining things: the Elasticsearch Python clients by day, urllib3 and trustme by night. I write about #Python, HTTP, async/await, open source and performance! he/him
81 posts
1,066 followers
183 following
Getting Started
Active Commenter
comment in response to
post
Probably not? But you can quickly switch between keyboards with a single tap on the earth icon. If you group langages wisely that could work well.
comment in response to
post
I did that initially but it sends the whole prompt again which is costly if there’s no context caching. It also anthropomorphizes LLMs needlessly.
It’s much better to evaluate the response with a thumbs up or down!
comment in response to
post
Yes! Which is why Seth switched to a faster but still mostly correct implementation at the time.
comment in response to
post
By the way for a few weeks in 2019, urllib3 used rfc3986 but had to stop due to the slow import times on Raspberry Pi with "only" 1GB of RAM: github.com/urllib3/urll...
comment in response to
post
It’s heavily text-based (a fancy dictionary, essentially) so doing that with Python (and migrating from 2 to 3) would have been basically impossible.
(Also I am not the original author of the app.)
comment in response to
post
Very cool! Upgrading from PHP 5.6 to 8.x was easy enough, I had to upgrade the framework I use, migrate from mysql to mysqli (by using the framework more) and stop using unset variables and unset array keys thanks to isset().
comment in response to
post
Nice! How do you track the time?
comment in response to
post
I would guess en.m.wikipedia.org/wiki/U5_(Ber...
comment in response to
post
It does not depend because you live in the US, of course!
comment in response to
post
Sorry if I don't get the joke, but no, it's @aoc.bsky.social, @georgetakei.bsky.social, and @markhamillofficial.bsky.social.
comment in response to
post
It’s a very difficult book to read in my experience. Many things don’t make sense if you haven’t read the short stories first. Many of those terms are used without being defined first: www.litcharts.com/lit/neuroman...
comment in response to
post
Merci pour cette explication accessible ! Et c'est toujours un plaisir de voir une collaboration imprévue entre deux chercheurs.
comment in response to
post
And to be fair, Starter Packs really helped here.
comment in response to
post
Ah j’avais pas vu désolé. Ils font un travail remarquable sur la modération. Je viens d’apprendre que les quote tweets pouvaient être détachés aussi : bsky.social/about/blog/0...
comment in response to
post
Ah dans mon cas c’était pas politique, juste la critique du langage Python dans l’espoir d’obtenir des followers. Mais oui ça existe d’autant plus pour de la propagande organisée
comment in response to
post
Et ça masque aussi pour les autres ?
comment in response to
post
Pour les postes en anglais il y a aussi des comptes qui répondent automatiquement avec de l’IA. Ça me dépasse qu’on puisse automatiser les commentaires désobligeants.
comment in response to
post
Ah je savais pas que ça bloquait pour les autres aussi, c’est top ! Un inconvénient c’est que la liste des blocs est publique : clearsky.app/pyviv.bsky.s...
comment in response to
post
I considered saying PRs welcome but I know you would likely do it 😅 And I am not sure if it would be a maintenance burden. I am hoping cryptography will work soon enough.
comment in response to
post
Neat! How do you generate the screenshots? They look great
comment in response to
post
Still, LLMs are a great way to learn Rust. Even Claude Haiku can output working code and explain basic concepts. For example, it tells you to call `.iter()` before zipping two `Vec`s.
(I also reference the public BurntSushi API a lot, specifically blog.burntsushi.net/rust-error-h... is 🧑🍳😘.)
comment in response to
post
The original solutions work with longer strings like say 01010, 0000, or 1110101. No just 0 or 1. Right?
comment in response to
post
Makes sense 👍
comment in response to
post
I also got the Yeti one step closer! Is this going to be only JavaScript? The home page doesn’t say.
comment in response to
post
I have heard good things about github.com/nektos/act 🤷
comment in response to
post
The details I was thinking about are in a GitHub comment instead: github.com/python/steer...
comment in response to
post
Yes, it is: docs.python.org/3/glossary.h...
There was also a Discuss post giving more details about the terminology but I can’t find it right now.
comment in response to
post
Thanks! urllib3 already received two PRs about this. Isn’t removing the classifier a bit early? PEP 639 is still provisional.
comment in response to
post
Same thing where I live! And since it’s a quite religious region, It’s mostly prayers.
comment in response to
post
Key Customer Metrics
Total customer count with Annual Contract Value (ACV) greater than $100,000 was over 1,420 compared to over 1,370 in Q1 FY25, and over 1,220 in Q2 FY24
Source: www.businesswire.com/news/home/20...
(Stock is up 25% since those results were published)
comment in response to
post
Interestingly the number of customers paying $100K or more is a core metric at Elastic!
comment in response to
post
So there’s no detector but we had a weird pytest-memray failure when upgrading to pytest 8 and @pablogsal.com helped us realize some server test threads were not terminated and memray was recording their allocations. Then @ecerulm.bsky.social fixed all such threads by hand.
comment in response to
post
It’s not a user-facing feature (urllib3 never spawns threads itself) but it made CI a lot more reliable: github.com/urllib3/urll...
comment in response to
post
Thanks to you!
And no more leaked threads! (With help from pytest 8)
comment in response to
post
Not really related to your point, but this specific block is likely not needed anymore. I have "Enable adult content" off (the current default) in the moderation settings and my Discover tab is a lot like my Following tab.
comment in response to
post
You can use @clearsky.app to find out on which lists you are
comment in response to
post
They were very visible here before the latest wave
comment in response to
post
Mastodon users were saying the same thing after the big wave in November 2022 ("10x less followers but 10x more replies!"). It’s easy when most of your users have just joined, they’re more likely to still be around. The challenge is to make this last.
comment in response to
post
Thank you! Also I’m happily following you as I have a NLP PhD, so our interests really align.
comment in response to
post
Hello! Since you mentioned self-nominations are OK, I maintain urllib3, trustme and work at Elastic as a Python maintainer. github.com/pquentin
comment in response to
post
A truly distributed team!