Just an idiot-level question: so these "constants" are subject to revision, but *how often*? What is the actual cost of recompiling and using them *as* constants, compared with the cost of rereading the stuff every time you run the program and passing it around? -- If stupidity were a crime, who'd 'scape hanging?

Richard A. O'Keefe writes:

Just an idiot-level question: so these "constants" are subject to revision, but *how often*? What is the actual cost of recompiling and using them *as* constants, compared with the cost of rereading the stuff every time you run the program and passing it around?

My apologies but I kind of lost track of this thread after the initial helpful replies and didn't follow up diligently on what went only to haskell-cafe (which I don't follow regularly) and not to my inbox. In case you are still curious an example of data that changes frequently is Earth orientation parameters (of which the leap seconds which Andrew elaborated on are one). These include e.g. the difference between UT1 and UTC (UT1-UTC) which is necessary to accurately relate the position and orientation of e.g. a ground based observer to an object in inertial space. UT1-UTC is continuously monitored with new observed an predicted values being published daily. Additional parameters and their update frequencies are listed at [1] and available compiled and with some documentation at [2]. [1] http://www.celestrak.com/SpaceData/EOP-format.asp [2] http://www.celestrak.com/SpaceData/ A piece of software precessing e.g. satellite orbit observations would ideally use the latest Earth orientation parameter data without requiring a recompile for each day. Thanks, Bjorn

On Wed, Aug 20, 2008 at 1:49 AM, Henning Thielemann wrote:

On Mon, 18 Aug 2008, ajb@spamcop.net wrote:

...

G'day all.

Quoting Bjorn Buckwalter :

...

I'd store the constants in a data structure along the lines of:

...

data AstroData a = AstroData { mu_Earth :: GravitationalParameter a , leapSeconds :: LeapSecondTable }

I would like to know if there is any consensus on what is the best way to make such a data structure accessible in pure functions. Passing it explicitly would be a mess.

In this situation, there isn't necessarily any shame in using a top-level unsafePerformIO as long as it's well-hidden:

module AstroData (AstroData(..), globalAstroData) where

data AstroData = AstroData Int

But here my argument about playing around with the Planck constant becomes relevant. _______________________________________________ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe

You can always change how you unsafePerformIO the data, though. If you want to set Planck's constant to 42 (or whatever), just change the unsafePerformIO $ <whatever> to unsafePerformIO $ return 42. Alex