Freigeben über

PDC05: Looking up

  • Artikel

As Emperor Zurg
might say, Los Angeles' power outage today
appeared to be caused by the high voltage equivalent of failing to connect the
plus wire to the plus side and the minus wire to the minus side. Anyway,
glad to see they got that fixed.

 

Had a nice dinner
tonight with a few VC++ team members and a few MVPs at Engine Co. No. 28. It's always good to
hear the issues faced by the MVPs, both in their day jobs and in their work
online with fellow VC++ developers. Conversation points
included...

  • One of the bigger
    issues discussed is the difficulty in upgrading compiler versions caused
    by stricter language standards conformance.
  • Performance of
    compiled code matters. This is especially important in the world of
    native-managed interop, where a variety of performance-sapping gremlins can
    bite you if you stray too far into the woods. Hmmm... too many
    metaphors for one sentence; I think it's near bed time. Anyway, note to
    self: the VC++ team needs to publish a white paper or two on /clr compilation,
    interop, and performance.
  • Jim Hogg and Brandon Bray gave me an
    education on the coolness of Haskell,
    which I'm now eager to try out.
  • I won $1 from April Reagan for remembering the quadratic formula.
    Truth be told, I was a little off, but is it really important whether it's 4ac
    - b^2 or b^2 - 4ac? :)
  • Brandon mentioned at dinner that today's precons went well,
    although they did suffer a bit of a delay due to the temporary lack of
    electricity. Apparently there was even some serious consideration of
    forging ahead with the presentation - GASP - without the benefit of the
    PowerPoint slides! Luckily the power came back up seconds before they
    descended into such madness. ;)
  • And I just noticed Raffeale already put a picture of dinner
    up on his blog. Man,
    he's fast.

Comments

  • Anonymous
    September 13, 2005
    > I won $1 from April Reagan for remembering the quadratic formula. Truth be told, I was a little off, but is it really important whether it's 4ac - b^2 or b^2 - 4ac?

    Hmmm, so is it really important whether the discriminant root gets multiplied by sqrt(-1)? Well, as the Bible reminds us, "If an 'i' offend thee, pluck it out!"
  • Anonymous
    September 14, 2005
    > One of the bigger issues discussed is the
    > difficulty in upgrading compiler versions
    > caused by stricter language standards
    > conformance.

    Aren't there already enough solutions on offer for that kind of person (project manager or whatever) who don't want to be informed when they're compiling their broken code?

    For example, they can continue using the previous version of Visual Studio which silently accepted broken code (including some syntax errors, for which the ISO standard requires diagnostics, but which an implementation is allowed to accept and compile as long as it has issued a warning).

    And/or, they can use #pragma to turn off an error message or maybe even turn off a warning (which helps make the implementation non-conforming, but if the user wants this then they get what they want).

    There is absolutely no reason for the default settings to remain silent when broken code is detected. Just teach those users how to change the settings. Don't make it worse for other users who want to do the right thing.
  • Anonymous
    September 15, 2005
    I think VC 2005 is pretty good about not remaining silent when broken code is detected -- I think we all agree this is a good thing. Most users employ the various backward compatibility switches and defines as a crutch to help migrate from earlier versions, but it still can be an effort, particularly for code bases that are large, old, or both.