IPv6, OpenBazaar, Swapping, JSP+, TLB, WLAN / WiFi

Post date: Nov 5, 2017 8:37:00 AM

• The register writes - No more IPv4. IPv4 address shortage is getting worse. No news? That's hardly unexpected. Yet making the news really funny is the fact, that the register doesn't yet support IPv6. This makes them awesome false prophet. Do as I say, not as I do. Hahah.
• OpenBazaar 2.0 is coming, with Tor support. That's awesome. Another cool feature is encrypted secure chat over IPFS and Tor.
• Windows Security Updates Guide (SUG) is here. No more traditional security Bulletins. No more patches, just monolithic 'updates' to the 'latest' version. No way to select what updates will be installed and which ones wont. Some friends hated this, but I can also see benefits. Having messy partially updated systems can be a source for many strange problems. If you've got single 'solid' version, it's much better.
• Julia Evans posts about swapping. These are the reasons why these topics are constant reason for argument. There's always lot of people whom don't know about these things and are learning about these. Some things might seem more surprising than others. Even if the logic behind the memory usage / allocation is totally well, logical. Things are weird and confusing, just because you don't understand those. But it's still a very nice post. I personally always recommend using swap, because reserving valuable ram with 'stale' stuff, just doesn't make sense.
• Just wondering when Suomi Communications / SuomiCom would provide IPv6 for VDSL2 customers. It seems that routing using 6to4 is extremely bad at times. I'm pretty sure that more attention has been paid to native IPv6 address routing, naturally.
• Can't stop loving cheap electronics. One led light power supply just short circuited and burned fuse. Well, that's actually ok. At least it burned the fuse and didn't actually start burning. That would have been way worse outcome.
• JPS+: Over 100x Faster than A* - Techniques: JPS+, "Avoid redundant paths on grids". Goal bounding, "Avoid wrong directions on any kind of map". NavMesh. Dijkstra. Sub Goal Bounding. Comments: Efficient route planning, is quite interesting science, because there's no single right solution. There are several separate approaches depending on multiple factors. Quickly looking the JPS+ Global Bounding seems extremely efficient. - Very nice examples. There are several ways to get the exactly same job done. Some are just bit more efficient than others. Also it's a big question if something can be pre-processed or not. Pre-processing is all again about different trade-off.
• TLB - Long discussions with fellow developers about TLB and problems caused by flushing cache with lookups. As well as MMU - Related performance issues.
• Had endlessly long and semi pointless discussion about WLAN / WiFi reliability and how much better 5 GHz is etc. Well, as said it depends. In some cases where 2.4 GHz is crowded and 5 GHz isn't, it might make big difference. On the other hand, if you've don't have good 5 GHz router with DFS/TPC you might be locked to use just four 20 MHz channels (Europa), which are probably getting quickly crowded. Also it's good to remember that there are serious signal penetration issues with 5 GHz. It's pro, it makes the range much shorter. But if people think they'll get better WiFi using 5 GHz when 2.4 GHz signal is weak, they're probably getting it totally wrong. - Some times people just got impression that something is greatly superior to something else, without considering the related facts. There are multiple factors to consider and overall consideration is only good way to go. Also had very long discussion about Fragmentation Threshold to split large packets smaller and usage of RTS / CTS etc. Truth is that those values are 'easy to tune' based on on-site tests. But there's no way to tell 'best settings' based on 'wifi is bad'. Duh! - Also it's easy to forget that the situation is dynamic, in changes all the time, so there's no best static configuration for any site.