Re: Limits on array dimensions

[Gregory Lypny]

Thanks.  I'm not a programmer, so please pardon my incorrect 
terminology.  I find arrays to be invaluable, especially when creating 
utilities that index large (300 MB and up) flat-file databases (what 
I'm doing).  However, once the number of array elements exceeds about 
150,000 (for example, to keep track of record ID's while indexing), 
processing gets progressively slower and eventually stalls even if the 
contents of the elements are small.  This does not appear to be a 
memory constraint.  My fix is to dump critical array variables to a 
file from time to time, delete the variable to free space, and recreate 
the array with new observations.  The difference in speed is remarkable.

	Greg

        Gregory Lypny
        Associate Professor
        Concordia University
        ___________________
        "Out of the blue,
          and into the black.
          You paid for this,
          but they gave you that."
		Neil Young

	http://pareto.concordia.ca



On Wednesday, July 30, 2003, at 12:02  PM, 
[EMAIL PROTECTED] wrote:

Message: 5
Date: Wed, 30 Jul 2003 08:22:36 +0800
Subject: Re: Limits on array dimensions
From: LiangTyan Fui <[EMAIL PROTECTED]>
To: <[EMAIL PROTECTED]>
Reply-To: [EMAIL PROTECTED]
On 7/30/03 1:11 AM, "Gregory Lypny" <[EMAIL PROTECTED]> wrote:

Wanna give me a hint as to which of the dozens of properties I should
be looking at?
No information on array, but it should be at least comparable to custom
property. So that gives 4 GB limit on what you can store in array, and
nothing else is using the same RAM, including other variables and apps.
Sorry, still no answer to your question. I assume 4 GB limit on the 
array
storage includes the overhead of array index (or a separated RAM space,
still limit by 4 GB).
Given all the limits MetaCard imposed on other objects, it is safe to 
assume
that the number of element (index) of an array is limited by the 
machine
memory, or up to 4 GB, if you have stored nothing in the array.

Don't take my words for it though. I'll be happy if someone else can 
prove
me wrong.


_______________________________________________
metacard mailing list
[EMAIL PROTECTED]
http://lists.runrev.com/mailman/listinfo/metacard