Ceph, Referrer, Spin, CSS, Webshop, WhatsApp, Joint Stereo

  • Even more servers crashing due to disk death with Ceph. This is getting kind of boring and annoying. Crashing systems can be interesting, until it becomes annoying, frustrating and boring.
  • Rechecked and configured Firefox referrer policy. Only send domain, when clicking on something. Without that referrer information is stripped. I know this breaks some sites, but so many sites are already broken anyway, that I couldn't care less.
  • Reminded my self about spin and graveyard spin - and PARE procedure as well as instrument flight and navigation.
  • Good talk about CSS Grids. Finally a way to get web page layout somewhat sanely defined. This reminds me from the XML flaws which I've seen. Some of the XML documents describe layout, not the data / content. Which AFAIK absolutely and completely defines purpose of using XML.
  • Tried to order stuff from two web shops, another was so broken, that their shopping basket didn't work, and another failed on payment. I guess they're making so much money, that they just don't care about user experience.
  • Good thing is that ScaleWay provides an API. Otherwise it would be way too annoying to launch new servers. Now I can just run a script, which tries launching new server(s) every 15 seconds, until is is successful. If the launch works out, then next call will follow immediately. 15 second delay is only applied, when launch fails. But in a way, this is way annoying. Even if it isn't a real problem for me.
  • Another usability thing about WhatsApp. Some people would like to be able to track the position in long message chain, where they're reading. So having movable last read pointer would be very nice. This applies to many other chat apps too. Yet another nice simple thing would be "play all" option. If you've got lets say 100 new audio message, you could just click it once, and it would play the audio clips as more or less continuous stream. Just like the good old phone answering machines did. But these seem to be way too big technical challenges for modern software engineers.
  • Even more great engineering. Samsung's Android update did break WiFi, now European channels like 12 and 13 can't be used anymore. I tried even taking SIM card out and re-inserting it. Nope, not working anymore. But yeah, these are things which seem to often be exceedingly complex things for many engineers to understand. ;)
  • Some web stores send continuous spam to your email, even if they do provide unsubscribe link. It's not working. Well done, good job.
  • In one project, lots of code is being written. And now I guess more code has been thrown out than is still being used. I just wonder, if there's any sane reason in these situations even trying to write a good code? Make some kind of experimental scratchpad alpha code, which does about what's required... Then if you won in the lottery, and the code will be actually used in production, slightly optimize and improve it and add bit of exception handling. Because doing that before you'll know the code will be actually needed, is just pointless. The client is changing requirements faster than development can write code for the project.
  • Laughed at one webcast where presenter's webcast contained popups from IM apps, which internal matters regarding other customers. Excellent job, this is exactly how it's supposed to be done. NOT.
  • Joint Stereo. Played with XXY oscilloscope and it's very clear why some music genres work better with joint stereo and with some types, it's practically totally useless. it also quite nicely reveals how loudness war - affects some music. There's absolutely huge difference between, real acoustic audio recordings and something which is totally virtually and artificially generated bit stream.