Blog‎ > ‎

OpenBazaar, Slack, SaaS, Python, Programming, SWIFT, SSD, Refactoring, Grabage I/O

posted Mar 11, 2018, 9:35 AM by Sami Lehtinen   [ updated Mar 11, 2018, 9:35 AM ]
  • A very well written post about OpenBazaar related risks. It's not all positive, but it's compact, honest and truthful. That's just the way I like articles to be. None of the things in the article raised, blaring bs alarms.
  • Had long discussion with one friend why Slack is so popular, if it's so bad and inefficient code.
    My conclusion was that. This is the constant trend of using cloud services. Slack is so popular, because it's SaaS. Running similar service, would be trivial. But nobody still bothers to. Many people talk about self hosted or distributed, etc solutions. Like I do. But truth is that people really do not want that at all. They don't really care, if the code is bad, or if it's performance sucks. As long as it just doe solve their problem.
    But this is actually good example, bad code doesn't actually matter at all. As long as it works, and works well enough, they're happy.
    This is also a perfect example why I'm using Python. Even if my code is 20x - 100x slower than actually truly well written code in most of cases it doesn't matter. Customers don't really care. My code works, and delivers what's required reliable. No, it isn't always pretty. It's usually just slightly optimized on and often robust only on critical parts . But it works very well, and I did't use excessive amount of time creating it. Also making it 'relatively cheap' for them.
  • Yet it's delicate balancing act. Because if the data mass grows, or the code isn't being run on servers which are 90%+ time on idle. Then the performance might start mattering. And that's where you should adapt. I don't even like keeping slack open, because it wastes so much resources. And if I'm annoyed about resource waste on desktop, think about mobile phones. Ouch!
  • Watched one documentary, which was really good. They were once prepared for staging, booby traps, honey pots and different kind of end games. As well as all kind of event triggers. Only thing they didn't mention but should have been obvious, is that most of this should be automated. Because computer systems and networks work fast. There's no time for anyone to do anything, when it's happening. If you're going to take over high value target, it should be as far fully automated as possible. As always, everything must be well rehearsed, tested and in this case automated based on script. It's not like, now we're in, what should we don next? You're out, before you even know it. On the other hand, with some operations covert indirect communication is the key to success.
  • US SWIFT hacking surprised people? Well, I was all the time thinking that they've got full access to the SWIFT data. It's just natural, it's so high value target. The full access can be of course obtained legally, nor not. But if it can't be done legally, than it needs to be done using some other means. Some information sources are just too valuable to be left untapped?
  • I'm still getting flame feedback about my post where I claimed that SSDs doesn't magically make fragmentation problem go away. I still completely stand behind my statements. Heavily fragmented files are slower to access and especially write on SSDs just as on other disks too. On some devices with large erase blocks, the situation can be radically worse. Which is proven by the experience. Just last week defragged lost of systems using SSD.
  • Usually every time I'm refactoring code, I reduce amount of technical debt. But the point was originally, to leave that debt intentionally. You'll never know if the project gets terminated, even before it sees light. So there's no point of making it too well. 'Working Proof of Concept' is enough, it doesn't need to be pretty.
  • One on-line game, one of my friends ask me to play. What's the sign you've been playing the game too long? Well, if you can maintain 50:1 kill ratio, maybe it's not fun anymore.
  • This is just so true when doing integrations. - Garbage in, garbage out - Yep. System integration won't somehow magically make your business work, if all data is ambiguous and processes are not being followed etc. Yet I've seen that being attempted several times.