Loading ...
Sorry, an error occurred while loading the content.

Re: [boost] PREPROCESSOR library is now in CVS

Expand Messages
  • Jens Maurer
    ... Both. However, I think compile time is more important, because we really don t want to clutter our regression tests with tests specifically crafted to
    Message 1 of 16 , Nov 30, 2001
    • 0 Attachment
      Beman Dawes wrote:
      > Are you talking about compile time, test program execution time, or both?

      Both. However, I think compile time is more important, because
      we really don't want to clutter our regression tests with
      tests specifically crafted to measure library execution time.

      I've added a CPU time output to all "execute" calls (this includes
      both compiler and test program execution), which shows up in the
      full regression test logs (Unix only for now). Anyone wanting a HTML
      page can distill the info from that.

      > Regardless, you might want to do some experimentation first to see if there
      > are meaningful differences between compilers.

      For the preprocessor library, there are huge differences:
      GCC is "immediate", Comeau chews along for > 1 minute or so.

      Jens Maurer
    • Beman Dawes
      ... both? ... Yes, agreed. If we do start testing execution time at some future date, it should be a completely separate set of test programs, crafted for the
      Message 2 of 16 , Dec 1, 2001
      • 0 Attachment
        At 04:24 PM 11/30/2001, Jens Maurer wrote:

        >Beman Dawes wrote:
        >> Are you talking about compile time, test program execution time, or
        both?
        >
        >Both. However, I think compile time is more important, because
        >we really don't want to clutter our regression tests with
        >tests specifically crafted to measure library execution time.

        Yes, agreed. If we do start testing execution time at some future date, it
        should be a completely separate set of test programs, crafted for the
        purpose, and compiled with optimization turned on.

        >I've added a CPU time output to all "execute" calls (this includes
        >both compiler and test program execution), which shows up in the
        >full regression test logs (Unix only for now). Anyone wanting a HTML
        >page can distill the info from that.

        Good.

        Hum... I tried to download the linux log, but got strange results. Claimed
        to download 2 megs in 1 second, and file was unusable. Tried again with
        same results.

        >> are meaningful differences between compilers.
        >
        >For the preprocessor library, there are huge differences:
        >GCC is "immediate", Comeau chews along for > 1 minute or so.

        Interesting. I'm looking forward to seeing the results.

        Thanks,

        --Beman
      Your message has been successfully submitted and would be delivered to recipients shortly.