I feel like it's more nuanced than OP writes. Presumably every log line comes from something like a try/catch. An edge case was identified, and the code did something differently.
Did it do what it was supposed to do, but in a different way or defer for retrying later? Then WARN.
Did it fail to do what it needed to do? ERROR
Did it do what it needed to do in the normal way because it was totally recoverable? INFO
Did data get destroyed in the process? FATAL
It should be about what the result was, not who will fix it or how. Because that might change over time.
> Did it do what it was supposed to do, but in a different way or defer for retrying later? Then WARN.
> Did it fail to do what it needed to do? ERROR
> Did it do what it needed to do in the normal way because it was totally recoverable? INFO
We have a web-facing system (it uses a custom request-response protocol on top of Websocket... it's an old system) that users are routinely trying to, ahem, automate even though it's technically against ToS but hey, as long as we don't catch them? Anyway, it's quite often to see user connections that send malformed commands and then get disconnected after we send them a critical_error/protocol_error message — we do have quite extensive validation logic for user commands.
So, how should such errors be logged in your opinion? I know that we originally logged them as errors but very quickly changed to warnings, and precisely for the reasons outlined in TFA: if some kewl haxxor can't figure out how to quote strings in JSON, it's not really something we can't fix. We probably should keep the records, just to know that "oh, some script kiddie was trying to hack us during that time period" but nothing more than that; it definitely doesn't warrant the "hey, there are too many errors in sfo2 location, please take a look" summons at 3:00 AM from the ops team.
> The problem with durable execution is that your entire workflow still needs to be idempotent.
Yes, but what that means depends on your durability framework. For example, the one that my company makes can use the same database for both durability and application data, so updates to application data can be wrapped in the same database transaction as the durability update. This means "the work" isn't done unless "recording the work" is also done. It also means they can be undone together.
A lot of work can be wrapped inside a database transaction, but never everything. You're always going to want to interact with external APIs eventually.
Yes of course. External calls still need to be idempotent. But the point is some frameworks allow you to make some, or even most, of your work safe for durable execution by default.
No it's different. Idempotent would mean that it can be replayed with no effect. What I'm saying is that this guarantees exactly once execution, taking advantage of the database transactions that make multiple data updates idempotent together.
You may have thought this is an objective observation.
For anyone over the age of 16 this comment is a loud expression of your political views.
Also, I find Reddit to be super funny. Just yesterday someone posted a photo of their brain MRI showing a tumor the size of a tennis ball and everyone, including the OP were having a great time.
> For anyone over the age of 16 this comment is a loud expression of your political views.
I'm curious as to what you think those political views are, because I strongly agree with what sidcool wrote (even if they didn't mean it the way I interpreted it) and I disagree with you.
I think that Reddit "is a biased cesspool of partisanship", but very much in both directions. Many subreddits are so wholly hard right or hard left that I think they're almost caricatures of themselves. And even for subreddits without a hard political bent, they are often the very definition of an echo chamber - they are great places to go where you want everyone to agree with you and you can see people who disagree with you get downvoted to oblivion. And, importantly, this is literally by design based on how subreddits are created and moderated.
I have rarely (not never, but rarely) made a comment that took a somewhat nuanced opinion where I wasn't heavily downvoted. And, contrarily, I have made similar comments on HN where, if I wasn't particularly upvoted, I received what felt like fair dialogue and back-and-forth with other commenters.
All that said, I still use Reddit frequently and find it frequently interesting, sometimes informative, and often pretty hilarious.
You really don’t know adherents of which political stance are constantly complaining about “lack of a sense of humor”?
Or who has been complaining about the “tech bias”?
I am glad if you were somehow not exposed.
I don’t find it partisan. I come across a lot of criticism of Democrats. And the current administration.
Also, if you’re claiming subreddits are partisan in the way mods want it to be, we can’t conclude Reddit as a whole is partisan, can we?
I sleep better at night thinking it is just a battleground of astroturfing bots fighting each other (at least on the main pages).
Everything from massive Russian state-actor bot farms testing newly trained LLMs popping out AI-generated meme formats before deploying domestically unknowingly getting into arguments with Israeli bot farms trying to raise support for some new movie series that will enable them to raise money for their next missile strike competing for eyeballs/attention from some uni student in a dorm room paying mid-sized black market companies in India to post comments telling you that cast-iron pans are too hard to clean so you should buy the non-sticks you saw on instagram (which are just marketing dropshippers in the USA selling the QA rejected pans from established brands).
Hacker News is Reddit with a nuclear downvote button and tone policing.
It's not that much better in terms of "dead internet," the bots are just more eloquent. In some ways the HN flavor of gamified engagement actively encourages worse outcomes than Reddit.
> Given there are something like 20 million people in the US illegally
You should probably check your sources on that one. That number is highly suspect and came from someone who was trying to sell you on upping budgets for ICE.
> seems counter productive to count people who aren't citizens in a article primarily about US health statistics.
The Claude MD is like the documentation you hand to a new engineer on your team that explains details about your code that they wouldn't otherwise know. It's not bad to need one.
Consider that every time you start a session with Claude Code. It's effectively a new engineer. The system doesn't learn like a real person does, so for it to improve over time you need to manually record the insights that for a normal human would be integrated by the natural learning process.
Yes, that's exactly the problem. There's good reasons why any particular team doesn't onboard new engineers each day, going all the way back to Fred Brooks and "adding more people to a late project makes it later".
I certainly could be updating the documentation for new devs very frequently - the problem with devs is that they don't bother reading the documentation.
If you are consistent with how you do your projects you shouldn't need to update CLAUDE.md nearly every day. Early on, I was adjusting it nearly every day for maybe a couple of projects but now I have very little need to make any adjustments.
Often the challenge is users aren't interacting with Claude Code about their rules file. If Claude Code doesn't seem to be working with you ask it why it ignore a rule. Often times it provides very useful feedback to adjust the rules and no longer violate them.
Another piece of advice I can give is to clear your context window often! Early in my start in this I was letting the context window auto compact but this is bad! Your model is it's freshest and "smartest" when it has a fresh context window.
I have a friend attempting to solve this. He's basically creating oauth for age verification. You sign up with his service and verify your age. After that it works similarly to oauth, but instead of return your identity, it just returns your age.
It's not a perfect solution, as he would still know who you are, but it's built in a way where you create a token locally to pass to the site that includes your age, and that site passes it to his site, which verifies the signature. So he knows who you are but not what sites you visit, and the sites know your age but not who you are.
When I read the title I hoped that EFF was going to do exactly that.
There’s also a way to improve it: Sell “age verification cards” in physical stores. Just like they are verifying that minors aren’t buying alcohol or cigarettes, they can verify that these cards are bought only by grown ups. Sure wouldn’t be perfect but it greatly improves anonymity and especially in paid-for adult services can be used as a payment method so repeat verification will happen for top ups.
It should be noted that Don Jr. is one of the investors who will benefit greatly if/when this goes through.
reply