realbasic-nug
[Top] [All Lists]

Re: Intel vs. PPC math gotcha

To: REALbasic NUG <realbasic-nug at lists dot realsoftware dot com>
Subject: Re: Intel vs. PPC math gotcha
From: Frank Condello <developer at chaoticbox dot com>
Date: Thu, 31 May 2007 02:15:16 -0400
Delivered-to: listarchive at realsoftware dot com
Delivered-to: realbasic-nug at lists dot realsoftware dot com
References: <fed2e2560705301533t6d8f66cdw86e8a3125dcf4128 at mail dot gmail dot com> <15F7B30A-E60E-4AA0-AF1E-F48054C42201 at great-white-software dot com> <5d7ca37e0705301827k7040dabvb66af85f6972cd70 at mail dot gmail dot com>
See the Apple docs:
<http://tinyurl.com/yqgeds>

Floating point comparisons are typically only useful when you account  
for some amount of error - i.e. check if the numbers are within +- 
some_small_number. Singles have long been neglected in RB however, so  
there might be some things RS can do to minimize these sorts of  
problems. The ability to define 4-byte floating point constants and  
some smarter compiling to eliminate conversions to/from doubles would  
be a nice start, but I doubt there's a sane "fix" for this. Compiler  
warnings are a reasonable approach, but RS hasn't been too keen on  
those in the past.

Frank.
<http://developer.chaoticbox.com/>

On 30-May-07, at 9:27 PM, Mathieu Langlois wrote:

> Sounds to me like an architectural difference between the 2 cpus.   
> I doubt
> RS can do much about it.  I guess we'll have to wait for them to  
> evaluate
> it.
>
> Math
>
> On 5/30/07, Norman Palardy <npalardy at great-white-software dot com> wrote:
>>
>> Sounds like it's worth a bug report either way
>>
>

_______________________________________________
Unsubscribe or switch delivery mode:
<http://www.realsoftware.com/support/listmanager/>

Search the archives:
<http://support.realsoftware.com/listarchives/lists.html>


<Prev in Thread] Current Thread [Next in Thread>