mastodon 4.7.2I have various notes about the French administration.
I also write code and stuff.
RE: https://infosec.exchange/@mttaggart/116518022621367937
Cloudflare doing shit like this is the perfect analogy for why cybersecurity is such a shitshow. They even offered their DDoS mitigation services to canonical. They could just like. take down the stresser service, but that doesn't result in money. I'm so sick of this "its acceptable to be on both sides because we like money." travesty we're forced to deal with on a daily basis.
RE: https://infosec.exchange/@lerg/116524161473318491
To every techie lolsobing at "HR is writing code, we're all gonna die": maybe we should have thought about it before now. We've been fighting tooth and nails since forever to make sure that there would be no regulation whatsoever on writing software. So wen we do a terrible job, it doesn't have any bad repercussion (on us).
The consequence is that for everyone else, software is kinda shit, and more importantly, doesn't matters. So yes, why not cut the expensive, loud, and obnoxious middle man?
maybe pushing code as quickly as possible with zero domain knowlege and ignoring all existing research is not the best idea after all it's hard to tell but clearly there is not way to know
well, at least, if your bank website is responding right now, it's probably really your bank. #cloudflare
Like seriously, if the attack was:
people need to be fired.
And that includes anyone supposed to do any security related stuff there. It should have been absolutely obvious in the logs.
Just heard Michael Moore (Anthropic) on Planet Money saying that with their cyber cyber tools and best practices, they can *assure* to fine all the bug before the code goes to prod. I'm looking forward seeing them being sued because they didn't.
So. Hear me out. The existence of a "dignified transfer" implies an "undignified transfer", where they kick the coffin down the staircase. I don't make the rules.
I find it so absolutely hilarious and on brand that the whistleblower in the Edouard Philippe embezzelment case is franco-german.
RE: https://infosec.exchange/@lcamtuf/116517194178120536
This, but on everything, with whichever AI gave free token that day, and nobody bother paying for an audit.
That's software dev right now.
OH yesterday from a non-tech person, at a tech event: "I was at the local chapter of the MEDEF (french bosses union) and we did a poll with our member asking if using AI increased their benefits, and ~70% said no."
With a very strong undertone of "WTF nerds?"
RE: https://indieweb.social/@Outpost/116251600317479260
so hmmm maybe people take note.
Todays cursed mimetype:
text/plain; charset=gzip+enc
wat
In todays' AI PRs fun: "here is a thing, I tested it against mock data, so you should test it against real data before merging it."
The non-mock data is public (the mocked data isn't, amusingly enough), it is just that they cannot be arsed to do more than the prompt.
(the code is simple, and by looking at it, I'm pretty sure it will most probably work, but folks, at least try)
Bon, la TBM vous êtes sympa, mais 3 interruptions de traffic sur le tram sur les 4 derniers jours, ça va pas le faire. Sans compter l'augmentation des tarifs, et avec la fermeture du pont de pierre tout l'été...
Not sure if it is a bug, but surely it is confusing.
If you run tor and require a specific country code for the exit node, with strict node to 1 (--ExitNodes {} --StrictNodes 1) it will properly raise an error if the CC in invalid (say FOO), but if the CC exists *and* and there are no exits in that country, it will simply fallback to whatever, without any warning.
