Announcement

Collapse
No announcement yet.

There's A Kernel Subsystem Being Dropped In Linux 5.3 As Easier To Start Over Than Fix

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • HadrienG
    replied
    Originally posted by andresdju View Post
    I can imagine 22nd century history books: "In 2012 CERN discovered the Higgs boson at the Large Hadron Collider (LHC), but that was shadowed by the used of an ugly Linux kernel subsystem to interface with their FPGAs and a patched obsolete CentOS distribution".
    Please don't joke about the psychological bias that leads people to tolerate the most terrible things at the infrastructure level as long as the high-level situation looks reasonable from their side. It's a serious issue, which down the line leads to giant unmaintainable codebases (because implementing new features is more important than fixing the existing code), and an inability to convince the politics to hire anyone to take care of the mess (because that task does not look sexy enough)...

    Leave a comment:


  • andresdju
    replied
    Originally posted by HadrienG View Post
    Turd polishing is pretty much the definition of CERN's system infrastructure. The entire LHC software stack is built on top of the technologically obsolete CentOS distribution by recompiling a more recent version of almost every single package from source using a CMake-generated makefile...
    I can imagine 22nd century history books: "In 2012 CERN discovered the Higgs boson at the Large Hadron Collider (LHC), but that was shadowed by the used of an ugly Linux kernel subsystem to interface with their FPGAs and a patched obsolete CentOS distribution".

    Leave a comment:


  • HadrienG
    replied
    Originally posted by Sonadow View Post
    Wait.

    CERN developed the FMC subsystem for the Linux kernel with other authors.

    "The authors" now claim that it's turd and removed it from their development tree, and have requested upstream to follow though with its removal in 5.2.

    So CERN developed turd for seven years before deciding that it's turd?
    Turd polishing is pretty much the definition of CERN's system infrastructure. The entire LHC software stack is built on top of the technologically obsolete CentOS distribution by recompiling a more recent version of almost every single package from source using a CMake-generated makefile...

    Leave a comment:


  • pininety
    replied
    Originally posted by Sonadow View Post
    Wait.

    CERN developed the FMC subsystem for the Linux kernel with other authors.

    "The authors" now claim that it's turd and removed it from their development tree, and have requested upstream to follow though with its removal in 5.2.

    So CERN developed turd for seven years before deciding that it's turd?
    Its not that bad. The problem is that most projects in physics do not see the need for good programmers (which is also not part of most physicists training, I kid you not, when I did my masters in 2016, I attended a "learn structured programming course" which was basically a fortran tutorial. After the semester most people had barely made it to write a small calculator program) and just let physicists do it instead. Which is a horrible idea because most of us lack the training to do this. I mean, just think about how insane you must be to invent a C interpreter, then just treat pointers and objects the same and then expect that people are actually learning to write proper C++ code if they do anything with this system. Icing on the cake, give it a ungoogleable name "ROOT" so finding anything for your specific problem becomes super hard. Then tell people they should always use this system because it would be a good "Industry skill". Yeah, my ass, lol.

    But this is really not a CERN problem, it is just that a lot of physics is done at CERN. It is a modern physics problem partially due to bad training of physicists and also a bad mindset of many physicists ("I do not need to learn how to actually program! I am doing science and as long as it compiles it is fine! Stop wasting time debugging stuff or writing actually nice code")
    Every project I have worked in so far had this problem. Luckily, people start to fight it but getting funding for full time programmers instead of physicists is also hard.

    Leave a comment:


  • CochainComplex
    replied
    Originally posted by Haxk20 View Post

    I woudlnt say they are wasting money. Every test they do we learn something new we didnt know before. They are doing good work. And as with NASA they get maybe 1% of public money which yes is a lot but still less then most people expect they are getting and they were able to prove Higgs Boson exsistence. But thats not the point. They are still only people and they can make mistakes too.
    Dont listen to the Wizard of Oz ...he is ranting on CERN. Dont feed the troll.

    Leave a comment:


  • starshipeleven
    replied
    Originally posted by wizard69 View Post
    Whatever it is, I just see this as further evidence of poor management at CERN! I've come to believe that CERN is best at wasting public money with actual science a distant second.
    Sorry what? CERN is doing science. As long as the code works and allows communication to the FPGAs they are perfectly fine.

    Leave a comment:


  • Haxk20
    replied
    Originally posted by wizard69 View Post

    Whatever it is, I just see this as further evidence of poor management at CERN! I've come to believe that CERN is best at wasting public money with actual science a distant second.
    I woudlnt say they are wasting money. Every test they do we learn something new we didnt know before. They are doing good work. And as with NASA they get maybe 1% of public money which yes is a lot but still less then most people expect they are getting and they were able to prove Higgs Boson exsistence. But thats not the point. They are still only people and they can make mistakes too.

    Leave a comment:


  • wizard69
    replied
    Originally posted by Sonadow View Post
    Wait.

    CERN developed the FMC subsystem for the Linux kernel with other authors.

    "The authors" now claim that it's turd and removed it from their development tree, and have requested upstream to follow though with its removal in 5.2.

    So CERN developed turd for seven years before deciding that it's turd?
    Whatever it is, I just see this as further evidence of poor management at CERN! I've come to believe that CERN is best at wasting public money with actual science a distant second.

    Leave a comment:


  • starshipeleven
    replied
    Originally posted by Sonadow View Post
    So CERN developed turd for seven years before deciding that it's turd?
    I don't see that as an issue.

    Leave a comment:


  • down1
    replied
    Originally posted by cl333r View Post
    Idk, but I'm fine with it as long as their hadron collider doesn't explode. Only goes to show they're just people and they can have shitty logistics.
    Well we can't say from this article how shitty this is. CERN engineers are there to keep their equipment working why working within their limitations, not necessarily to reach computer science perfection.

    Leave a comment:

Working...
X