Fingerprint, Broken World, Usability, Duplicati, Standards, Data Quality

  • One thing about the endless public key fingerprint vs "trusted certified signed key" haha, aka trust chain signed key. I just wonder does the people claiming that signed key is safer realize that the signature only relates to the fingerprint. So if you know that the fingerprint is right, it's still better than trusting the (technically) broken certificate trust chain stuff.
  • About broken standards and processes. What's the difference between working vs documented vs specified software / process evolution of disinformation in chain. What if the original specification is just fine. But then there's some example code, which is totally broken or just badly implemented. When other implementations are tested and or created based on that sample implementation, does it mean that the protocol is broken? It's not if it's done as specified, but it might well be practically broken. Similar issues arise with piece meal project negotiations. There are several small pieces which are sane on some level, but when you combine all of those pieces the end result is totally broken and almost insanely bad. - I guess this is something which is happening in almost all large projects and or standard specification groups, etc. Because most of the people working on the stuff, doesn't understand the whole, nor care about it. They just push their own agendas forward.
  • More rants about usability? One banking application says, that you've got authentication / confirmation request. When I click, it asks for PIN code. But in around 2-3 seconds, the form resets and it requests the same information again. I think that's utterly lame engineering and design. If it needs to retrieve some additional information, how about NOT showing the PIN code request and telling user to wait. And if additional information isn't required, so why is the form refreshed by resetting it in a few seconds? Anyway, either of those two options would be just fine, but they've chosen to go between running the user interface and showing their utter incompetence. Good work, keep it up. These are just the things which easily annoy users.
  • Confirmed that Duplicati will use the --tempdir parameter for temp files, even if separate asynchronous-upload-folder parameter isn't being used, which is what I assumed. But you'll never know without testing. Yet later found out that tempdir parameter isn't being used with Linux, when testing or repairing archives. Only environment variable option will work.
  • Fixed some statistics collection code, which was quite broken. Business as usual.
  • Standards and specifications and documentation vs reality. The second bullet on this post. Today it happened again. Organization A complained that process B isn't working as documented in document C. Well of course it isn't. Because the organization A misimplemented it on based documentation C and because for their incompetent and lousy organization it was too hard to make that change immediately to fix it, we adapted and implemented the code so that it would work as they've implemented it on their side. Now after several years, they whined that it's not working as documented. Of course it isn't. Because you've changed the process and documentation hasn't been updated since. It's a good question why they didn't make the change immediately when it was obvious that they've misimplemented something. - Well well, business as usual. Just as depressing and ridiculous as usual. In some cases it's just amazing how hard some extremely simple things are, when there's enough junk organization and experts involved. If there would be just two competent guys, the stuff would be agreed and completed in a few hours. But when you involve endless and pointless meetings, disorganized experts and some political discussion. Whoa. 1000x resources consumed, and end results is substantially worse that it would have been without the bunch of overhead dudes. - Great work!
  • Data quality, endless discussion about extremely bad data quality, which is practically utterly useless. Classic $hit in, $hit out. Then people think that getting new data collection would solve the problem. Nope, it doesn't. If the users don't give a bleep, whatever system you're using, the data is still going to be totally useless. - I think it's funny and sad to have these discussions over and over again. Always the same thing. Duh! - This has been quite a fun week (including last weekend), lots of tiringly common and unfortunately very classic issues.