Negative, You, Login, H3/OE, TB, Prequal, LE, Btrfs

1. Someone pointed out that I've got so many negative views of different things. Sure I do, some things are just bad. There are many services and platforms, applications and solutions which are so good that it's very hard or almost impossible to come up with any serious complaints. So, it's not like all of my views are negative. I truly love and enjoy some solutions and some just work so so great, that there's nothing negative to say about those.

2. You.com AI site gets another honorable mention of providing excellent UX. Image generation has been broken for months and still is, and they ignore the issue and won't fix it, even if it would be trivial. Any person whom would give even a slight brown stuff, would be able to fix it. No, they don't even need to be computers specialists. If they would be web developers, it should take less than 5 minutes to fix it. - So classic.

3. Whoa, it seems that Microsoft n00bs have finally fixed their thought free login redirect loops with copilot. Now the login and redirects actually, instead of leading to infinite redirect loop until some rate limit cuts off the endless redirections after a while.

4. It's so nice that HTTP/3 (H3) offers Opportunistic Encryption (OE) (@ Wikipedia). I've encountered so many situations and cases where constant problems and issues with certificates have lead to opting for plaintext because encryption is just broken. This disparity has annoyed me for a long time, finally there's sane solution which allows protection against passive snooping.

5. Thunderbird, did I say it's enraging at times? Now it claims that I don't have permissions to compact RSS Feeds folders. Bleep, I can see form time stamps that the folders and been accessed and used today, and all permissions are ok. Yet, as usual, the error message doesn't specify which path it's whining about. I'm pretty sure it's not Feeds and it's files. Classic, crap code, omfg. It might be whining about temp? Maybe? But if so, why it doesn't then say so? - Checked Error Console and Console, nothing at all. Great, absolutely amazing. Well, I can compact the folders by creating new one and moving content to the new folder. But it's amazing that the developers can't do that. - GG - Based on my general experience and this situation, it seems that the compaction code is just crappy and buggy. - At least Thunderbird doesn't anymore constantly corrupt the cursed Mork files as it did at one point, maybe we can call that a success?

6. Load is not what you should balance: Introducing Prequal (@ YouTube) a NSDI'24 (USENIX) talk. Classic load balancing task, let's see what they do so much differently, because they claim this is way better than other similar classic solutions. Probes, Requests-in-Flight, Hot / Cold Servers, Probe Pooling, Remove Worst from Pool. WRR vs Prequal statistics, looks good. Final conclusion, you should minimize latency not load. This is just one way of doing it. And as far as I found out, there was nothing special or new about it. Similar results can be achieved using so many different methods. Happy that the included comparison with other policies at the end of the talk. - Also handling failure modes would have been interesting. It's classic trap that some node stops working properly and suddenly it seems to perform very well, because it's not actually doing it's job. Minimal latency, high RPS throughput. Yeah, this one (or a few) servers can now handle everything. Oops.

7. Let's Encrypt and Short Term TLS Certificates (@ letsencrypt.org) - Yup, certificates which are valid for just a few days or hours? Well, why not. I personally still miss something like actually working PAKE. In some cases TLS and certificate stuff works just so badly, it's easier to drop the whole encryption. At least H3 is bring in it the OE. - Sigh.

8. Some things are quite bit different with btrfs, spent a enjoyable while while configuring swapfile [root @ subvolume, no compression, no copy on write] and .Trash-User folders for Thunar, oh joy. But got it done. Newer versions of btrfs have a separate create swapfile feature: btrfs filesystem mkswapfile [-s size] file

2025-07-13