przemek0

's profile
image
engineer

Biography has not been added


przemek0

's contributions
  • 07.18.2017
  • Catenary feed seems like it can’t possibly work, yet it does
  • It probably helps that the train is moving: running 300A across this flimsy contact leads to arcing that you can literally see when observing moving electric trains---but the arc moves along with the train and so it does not burn through the contact wire or the pantograph. Actually, if you look closely the contact wire tends to move around across the pantograph electrode as well, which equalizes the wear.
  • 05.26.2017
  • Microsoft’s hardware support burden
  • As a resident Linux stick in the mud, I would point out that Linux is pretty good at retaining support---once a device works, it tends to keep working. This is probably because people who originally used a given driver tend to raise ruckus if it breaks, and it's much easier in Linux to determine which update broke it, and revert or fix it. I find it curious that Microsoft doesn't do such testing internally, at least for their own hardware.
  • 09.21.2016
  • TI enters calculator market, September 21, 1972
  • it's crazy that you can buy an android tablet for $25 (e.g. http://www.mojosavings.com/craig-android-tablets-only-25-at-cvs-reg-99-99) but the TI graphing calculator used in schools is $120.
  • 08.31.2015
  • Enhance IoT security: use snipers
  • This reminds me of another 'stealth' security technology that is advertised as "you can't hack what you can't see". This is naive: after all, Stuxnet hopped an airgap. One has to be very careful not to fall into "security by obscurity" trap---there has to be a solid security infrastructure besides 'stealth'. Having said that, Haystack has a good point---minimizing traffic is a good idea from many points of view, from battery life to scalability to, indeed, aspects of security.
  • 05.28.2015
  • 10 must-have development tools that (mostly) won't strain the budget
  • Re. file comparison tools, there's a bunch of FOSS comparison tools I use often. I was surprised that some of my colleagues didn't know about them, perhaps because they aren't part of a standard tool set like git or eclipse. Specifically, tkDiff and beediff are very good at displaying side-by-side versions, and visually connecting common and differing parts. I even used word2text-fmt-beediff pipeline to compare and contrast the content of two Microsoft Word documents, as I couldn't find good Office tools for that.
  • 01.05.2015
  • Is an Open Source Operating System Really Free? Not by a Long Shot
  • The article focuses on cost, and misses the fundamental point that the OS is just a tool to implement the functional requirements. The purchase cost and licensing cost is just one of considerations, while NRE engineering development cost dominates most projects, especially in the complex IoT area. In such case, best strategy is to pick the toolset that your best developers are most familiar with. The article offers Windows Embedded as an alternative, but Windows has not been very successful in the embedded area---the widely deployed technologies that make Windows a popular desktop computing platform, such as VBE and .NET, are not good match for embedded systems. Even more importantly, driver support for esoteric hardware devices and protocols is just not that great in Windows. In contrast, Linux has an amazing breadth of support for all kinds of weird hardware, and by now there's a lot of engineering talent familiar with Linux and Android platform.
  • 04.15.2015
  • Drivers are driving me crazy
  • I am sorry to say that, but over the years I came to the opinion that HP just sucks at software. I used their stuff for 30 years, starting with HPUX, through various printer/scanner drivers and firmware, and experienced consistent problems and uncalled for bloat. Great hardware, though!
  • 04.10.2015
  • Promising hardware hampered by buggy software
  • Great, but the point of Open Source is the community collaboration and rapid development cycle. ASUS (and many other OEMS) may use FOSS as a base but they care neither about a viable development community around the project nor about collaborating with other OEMs on a clearly defined upstream . I believe that RMerl forked something that is supposed to track their firmware but it's just one of many dev streams, so that the improvements are diluted across them and don't migrate into a common upstream that benefits all of them. Contrast that with well-run projects with a clearly defined upstream path collecting and redistributing _all_ the good stuff.
  • 04.10.2015
  • Promising hardware hampered by buggy software
  • It's shocking how complex the firmware got, and small wonders that OEMS have hard time keeping their game up. It seems to me that a good alternative would be a common Open Source stack that everyone is working on---in short, OpenWRT. Many routers have already existing OpenWRT/Tomato ports, which tend to have good reputation. Is it an answer to everyone's problems? Of course not---setting it up requires serious research and commitment, but given the amount of time Brian spent messing with his OEM firmware he may have been ahead if he just went with OpenWRT from the beginning.
  • 03.13.2015
  • 10 tips for commenting C code
  • There's a system for writing code called Literate Programming. It takes the good advice from this article to the next level---for instance, the compilable program is derived from the documentation---so that you'd write documentation annotated with pieces of code, and then automatically generate two resulting artifacts: a printable, indexed docs and a compilable program. As you can see, LitProg is a little more complex and available tools feel like 90-s. Still, the underlying idea is very sound and I enjoyed using it for programming very much.