If we could go back to Google 2010 that would be fantastic.
608 karma · joined May 16, 2014
If we could go back to Google 2010 that would be fantastic.
I knew some of the engineers at Unisys who were still supporting Clearpath and Libra. Even they thought Burroughs was weird...
We’d spend 20x development time on kernel code because BSOD is never an option.
I get that this was a bad release - but IMHO it’s incredible that they pushed this out to a billion devices before the red flags went up.
We put too much code in kernel simply because it’s considered more elite than other software. It’s just dumb.
Also - if a driver is causing a crash MSFT should boot from the last known-good driver set so the install can be backed out later. Reboot loops are still the standard failure mode in driver development…
Nonce collision is a huge concern on large file system deployments. 2^32 seems huge but when you’re writing 100k iops a second on a PB array the chance of collision is almost guaranteed if you’re betting on PRNG randomness.
I was invited to the Atlanta Linux Symposium in the late 90s and met just about everyone around the Linux/Slashdot/X11 world. I had been hacking on OpenGL drivers and one of the first to "integrate" 3dfx 2d acceleration into X.
I had a wonky alpha AXP desktop I'd built from random parts and porting X11 to 64 bits on Linux as the code & binary was actually still 32bit. I'm chatting with rasterman for a bit and up walks an older gentleman.
He'd been listening in on our conversation and was eager to get my take on X. I told him straight-up I thought it was kinda crappy and it should have used blah blah bah. He said "I'm Jim Gettys and I co-wrote X windows.." I was aghast...Oh shit. He laughed and we had a long conversation about how X was implemented.
Shouldn’t model developers disclose their alignment training questions and censorship triggers?
I don’t disagree that there needs to be guardrails on dangerous information to a certain extent.
Companies producing these models often go beyond general safety and regurgitate political points as if they were the same as how-do I-make-nitroglycerin questions.
Hot political topics are most-often bucketed into a single ideology.
Disclosure would go a long way towards understanding the motivations and goals of these models.
Our job was to implement a crypto storage system for an OSS database where each user only had access to data based on their own authentication keys. Only minimal changes to the DB allowed, so we made our own PAM module to handle the authentication and key management.
We implemented a POSIX layer that intercepted file ops and backing-stored and encrypted the blocks into S3 containers. Files/blocks that got purged would automatically pull from S3.
One month into the project the major software company decided they didn’t want it. But we still had the contract and had to fulfill it for six more months.
So yeah, six months of code in the trash. But I learned a lot of S3, AWS and wrote a toy compiler with the spare time. Since nobody cared I was able to try a lot of new solutions, test out different languages and tools. Basically experimented with everything out there.
Delivered my code on the day of the deadline. Tested, worked, archived.
We tested cmpxchg16b and found the performance was terrible with more than 4 cores.
Ended up using spin-locks similar to Linux kernel RCU.
I'm guessing the initial puff creates a high pressure area on top of the paper, rolling it downward and back. Them after the puff has pushed the air away, there is now a low pressure zone on top of the paper which lifts it up as the air below is rushing upwards around the sides of the paper.
I had a threaded server that we were debugging which would only dump state correctly if we deleted a printf right before in a different thread. Really confused me until I figured this out.
I really like the concept - good execution. I picked up the ability to click on a word for translation easily.
It worked the first time after asking for mic permission.
But second time, I did have a problem with the mic button on iOS 16.6.1 getting stuck on but not taking audio input.
Pushing the button again (to cancel perhaps) did nothing.
Switching to tutor mode fixed it for another input.
I see there's "translate-c" to help migrate C to Zig, and maybe that's enough.
Generally I fall on the side of if a compiler can toss out a type warning, it could certainly give a reasonable solution also. Best-guess with a "// Warning type was guessed" is better than an opaque "RTFM" in most every case.
We researched ways to pay him legally, but for a six month contract it turned out to be much easier to simply pay an agency that pays him and does the paperwork.
Essentially it is presumed that most workers are full-time employees, and it is difficult to maintain a contractor due to tax liability and labor laws.
I used potrace and autotrace to darken the faded text strokes - I should give this a try and see if it can do a better job.
I highly recommend - best model we tried out of dozens of public models.
In NIST SP 800-90A Rev. 1, the HASH_DRBG section has been significantly updated to that effect.
For instance, Appendix E: (Informative) Revisions.
Section 10: Section 10 now includes a link to the DRBG test vectors on the NIST website. Sections 10.1, 10.1.1 and 10.1.2 now include short discussions about selecting hash functions to support the DRBG's intended security strength. The Dual_EC_DRBG has been removed, and section numbers adjusted accordingly.
For example, these devices were validated for the completely appropriately named "SonicOS 6.2.5 for TZ, SM and NSA". Gotta appreciate the irony.
Cavium CN7020 Hash DRBG
Cavium CN7130 Hash DRBG
Cavium Octeon Plus CN66XX Family Hash DRBG
Cavium Octeon Plus CN68XX Family Hash DRBG
I don't know if that's hardware support or just a software validation - but it's still interesting that they validated it.
https://csrc.nist.gov/Projects/Cryptographic-Algorithm-Valid...
From Wiki: >>> NIST SP 800-90A ("SP" stands for "special publication") is a publication by the National Institute of Standards and Technology with the title Recommendation for Random Number Generation Using Deterministic Random Bit Generators. The publication contains the specification for three allegedly cryptographically secure pseudorandom number generators for use in cryptography: Hash DRBG (based on hash functions), HMAC DRBG (based on HMAC), and CTR DRBG (based on block ciphers in counter mode). Earlier versions included a fourth generator, Dual_EC_DRBG (based on elliptic curve cryptography). Dual_EC_DRBG was later reported to probably contain a kleptographic backdoor inserted by the United States National Security Agency (NSA).
From Cavium's NIST FIPS-140-2, Section 3.3 [1] Approved and Allowed Algorithms:
The cryptographic module supports the following FIPS Approved algorithms.
*SP800-90 CTR DRBG Deterministic random number generation 32
1: https://csrc.nist.gov/csrc/media/projects/cryptographic-modu...
- Manufacturer voluntary guarantee of 1/3/5 years security updates with an expiration date.
- Separation of functionality and security updates.
- The ability to "turn off" connectivity and retain full local functionality.
- An industry security certification like UL.
- A single point way of identifying and validating devices.
As it is, I avoid using IoT mostly for security reasons. Having worked in security for many years I have seen the best and the worst. Having security isn't a panacea either - it needs an ongoing management & reporting infrastructure.
Agreed on the current state of the FDA filings - there's a lot of paperwork and process auditing - but guidance is lacking and I'd like to have more clarity rather than just "industry best-practices." That said - things are much better and it seems like the trajectory is improving.
Last week had to fix the instrument clusters, door handle, and a couple of recall issues. My 100k warranty ends in 400 miles - so needed to get it done.
This was my first time in the shop. Got a loaner for 4 days, problems all fixed, no cost. They communicated everything going on.
Really fantastic- I feel like a valued customer. Only thing I’d suggest is they wash and vacuum - but that’s just a nice-to-have really.