JavaScript, Fingerprints, Linux, Nvidia

  • Optimized one JavaScript program, which updated status document after every byte being uploaded. I changed the code so, that all key information is only updated once per second. On faster network connections the bad page update code caused CPU to be pinned running the bad JavaScript earlier than network got saturated. - This is perfect example of code that works, but is practically very bad. While fixing up, found out that the original source code was obviously horrible mashup of different code snippets from random sources. There were many global variables which weren't referenced at all. Some elements were updated which weren't present in the HTML document and so on. It was obvious that the code was really really bad. I just fixed the worst parts and removed invalid references. At least now the console doesn't get full of invalid reference log entries. - No, I didn't rewrite it all, which would have been the only sane action to take. But it wasn't worth of it, not this time. - Yes, there were also lot of unused gunk in CSS file, but I didn't bother touching it. I just added a not that I'm aware that it's junk, so nobody bothers to complain about it.
  • Believe it or not, even more continued discussion about if certificate fingerprints are valid and if or some other authority should be used and which one it should be. I guess next step is to take the standard step, and start accepting any public key / certificate for server and get rid of all the problems related to the matter. - Omg, now they're confusing client / user public / secret keys with server keys and so on. - My mind is melting. - I just wonder how hard and complicated simple things can be. - Perfect example of group of experts, which makes everything extremely slow, messy, hopeless ... Now they've used three days and had several meetings, about confirming one public key's fingerprints authenticity and confirming that it's ok. - Duh. - Is this the servers public key's fingerprint d4:f9:77:ff:73:84:02:ac:d6:6d:26:fe:fc:fd:e1:55 which we're supposed to send the confidential data? - Only Boolean answer is accepted: Yes / No? - Can't be that hard. - Then some user rights have been messed up, while figuring out this matter and ... Now this starts to sound like business as usual. Then messup about which user it is about and which domain. Trivial matter takes 20 top engineers and takes a week to do.
  • I think I've mentioned Linux display driver issues a few times. Well, I just got more of those. #Linux #display #drivers #broken again. Today I just noticed that Xubuntu and Lubuntu / Ubuntu are suddenly broken with latest kernel and drivers on a few computers. So annoying. I've already replaced one display adapter because the drivers just got so badly broken on Linux. Someone tried claiming that the display adapter was broken, no, that's not true. It worked perfectly with Windows. As well as with older Linux live boot. This seems to be quite common trend, how annoying. Even if I choose old enough #kernel from boot menu, the exactly same setup works, and that's the only change. Sigh. Current issue is with: Display controller: Intel Corporation Mobile 945GM/GMS/GME, 943/940GML Express Integrated Graphics Controller. Display memory data is clearly corrupted in somehow. I had similar problems earlier with Old Nvidia GeForce display adapter when used with Intel (Integrated i7) adapter. This issue happened already several years ago. I've replaced the display adapter since. In both of these cases the issue was exactly same. Memory buffer data totally corrupted. But annoyingly the mouse pointer works perfectly on top. Funnily on the current Intel case, the display gets only 4/5th parts corrupted. But not from top down, but from left to right. So the 4/5 parts of screen is corrupt but the 1/5 part of (right most) screen is working as expected. I guess it's some kind of quite simple stupid fail in the driver code.
  • When I started nagging about that. There's yet another problem with the current Nvidia driver. When it enters power saving mode and should turn of displays, all the screens just turn blank, with the back light being left on. So screens are blanked, but not powered off. I've tried all kind of settings and different power management stuff. But I guess it's the Nvidia junk code again, which is inherently broken. Also sometimes when seeking video with VLC display stop working completely. If there's some hardware video acceleration, it's probably buggy too. Yes, I can still SSH into the computer and everything works. But all displays are frozen and console mode etc won't work, because displays are screwed up, not only the X session. This issue occurs with NVIDIA Corporation Device 1c02 (rev a1).