NoScript, OwnCloud, AppIMage, IPsec, PEP 572, Outlook, BASE
- Studied NextCloud and ownCloud - NextCloud seems to be the better option of these two, clearly. After short tests, I decided that I don't need such solution after all. More or less basic Nginx + SSH works well enough and is simple to setup.
- Reminded my-self about AppImage. In some cases this is absolute perfection. Everything just works. It's so tiresome at times to configure software, which takes several days and even then, might end up in totally insecure configuration or break down repeatedly after updates and so on, managing dependencies, failing to compile and all the oh so familiar chores of system administration & software development.
- Really nice post about IPsec.
- Simple, correct, fast: in that order, yet I would place working (in practical terms) before the simplicity. Yet if possible, I would design the process so that correctness can be achieved as simply as possible. Of course there are cases, where you can't change the things, but in many of my cases I can, in others I can't. - Maybe one of the examples is like the Selective Reject, which makes re-transmission more efficient, by not re-transmitting received segments, but it also makes the process more complicated, than just saying please re-transmit since X. Many improvements just make things more and more complex. Another example of this is the UDF file system, every version is just more and more complex than the previous one to implement.
- Red flags signaling your rebuild will fail - It's easy to think that Rebuild will fix things. Devil is in the detail. Implementing something simple in more efficient way is quite trivial. But when all real world dependencies and requirements, aka complexity is present. It probably won't be that simple anymore. That's just like how to eventually consistent, document databases, are great solution for performance. As long as you don't need support for all the stuff you can do in the SQL. Similarly locking and parallel access is simple, as long as you keep the locking coarse enough. Which actually blocks the efficient parallel access completely. Or you can build layers on top of layers, but then the pile will be complex, even if each layer would be seemingly simple.
- PEP 572 - This is great simple example. I kind of like simple syntax. That's why I liked Java. It's syntax is technically simple. Yet it often did lead to very sparse code, which I do dislike. PEP 572 is kind of niche code compaction (?), and makes syntax more complex. But sure, it can save a bit of code in some places. I'm not afraid of a few extra lines, if alternative is more complex and "obfuscated" code. If obfuscated code isn't a problem, I would think everyone would be using regex and perl all the time. - Today I wrote some code, which would have cleaner code using this new structure.
- Outlook - Sick'n'tired, it's rejecting lots of email again. Just so annoying. It seems that spam protection is huge problem to Microsoft, they'll get it wrong, time after time and in so many different ways. It's totally enraging. Interestingly Outlook seems to be the only provider with such problems, for a long time I haven't seen any problems with any other providers.
- Long discussions about problem space vs solution space. It's important to describe the problem so that it doesn't force solution which isn't optimal. Yet this is of course very common issue when ever a problem is described.
- Basically Available, Soft state, Eventual consistency (BASE). Which provides better performance when sharding, because there are often major limitations to atomicity of data state. Which can be improved with Strong Eventual Consistency (SEC).
- Something different? Gabriel V anti-ship missiles, StarTram and yet another rocket free solution for space launch.