Matrix, Teams, EGNOS, AI, JTAG, JWST
Matrix Sliding Sync MSC3575 (@ GitHub) - Yeah, original protocol was really bad and naive. This one looks somewhat sane. Good.
Matrix Invisible Cryptography MSC4153 (@ GitHub) - Still many things I don't really like. Many things say should. As we know, developers ignore all that junk. If it's not MUST / MANDATORY then, well, it's not. - Pretty much perfectly corresponds to my comment about exceptions with cURL. When someone asks how it works, the only correct answer is, well it depends. Do users get warning? - Yes and No. If you say yes or no, you're wrong. To answer that question you'll need to list clients and versions and different conditions and... Yup...
Matrix Authenticated media MSC3916 (@ GitHub) - Only good thing about this, is that they acknowledge the problems. Same thing as with the Sliding Sync, initial implementation was really bad and naive. Good that the problems are getting addressed, even if not yet solved.
Microsoft Teams - Scheduled message sending is extremely badly implemented. Confusing, unreliable, broken, and bad UX. Good job guys! Sometimes the scheduled messages are just lost, and the app might show that you've got scheduled messages for this chat, but you can't find those anywhere, etc. Also sometimes messages assigned as scheduled are shown as sent, until next refresh and then you'll see that it's scheduled. Very basic mistakes and shows no attention to the task, nor any competence from developers. - I would understand it if it would be hacker news post, I'm 14 and I wrote my first chat GUI. But elite team of experts making this, bleep?
EGNOS (@ egnos.gsc-europa.eu) and European Geostationary Navigation Overlay Service (@ Wikipedia), seems awesome system and it's free for everyone to use. Ah, what GNOS is? Well it's WAAS, SBAS and also the Augmentation System data is openly available over the internet using EGNOS Open Service SDD. Providing decimeter or 10 cm navigation accuracy.
Absolutely great cipher machine information resource the Crypt Museum (@ cryptomuseum.com)
I've been wondering for a good while, why trim (discard) is so slow on some devices, but this helps (to know why): hdparm -I /dev/sda | grep -i trim - Yep, different devices support very different number of blocks to be discarded with a single call. As well as you can set the rate limits for trim. Some devices trim lot faster than others as well as of course the file system free space fragmentation level also affects that, because there might be more or less contiguous areas on the drive that can be trimmed as a single block.
Nemotron-4-340b model tested, works good. It's just strangely large compared to the size of the training data set they used. Makes me think more like hardware testing, instead of having small and as efficient and optimized as possible, aka distilled model.
Read a few documents about JTAG (@ Wikipedia) and SWD interfaces. Both got pros and cons. But at least currently I don't luckily have any devices I would need those to debug for.
JWST (@ Wikipedia) - Summary Documentary, watched a documentary summarizing the results of JWST achievements and observations for the last two years. - Amazing stuff, it's absolutely worth of the investment.
A friend installed Geotrim Trimnet VRS GNSS high accuracy receivers achieving centimeter (10 mm / 0.01 m accuracy).
PySkyWiFi (@ robertheaton.com) - Ahaha, anything over anything at it's best. Or well, technically just chunking and tunneling, but a good story.
Had a long chat with security & privacy people about Matrix and it's permanent record. True, it's yet another protocol, which doesn't handle data hygiene well at all. Users don't know when the data (and or keys for it) will be gone, if ever. Also the client currently unnecessarily keeps the keys for all messages, even if the messages have expired or been deleted, so there's no need to keep the keys. Based on good data hygiene standards, if something is unnecessary, then it should be deleted. There's always risk that the deleted messages are still stored somewhere and when the keys are kept forever, then it's possible to decrypt those messages with the keys.
SimpleX Chat - Option to backup just contacts is still missing. From data hygiene point of view, it's not a great idea to backup everything. But these are hard concepts to understand for many people.
2025-01-26