"Dumb question" time

Richard Gaskin ambassador at fourthworld.com
Wed Jul 27 14:17:22 EDT 2005


jbv wrote:
>>The only issue I've ever seen with the sort command turned out to be
>>documented:  the string library used for it requires that none of the
>>lines exceeds 64k (65,535 characters).  When sorting lines in which one
>>or more is longer than 64k the results will be unpredictable and likely
>>confusing.  I know, I've seen 'em -- sure confused the heck out of me
>>until I learned this limit. :) 
> 
> Thanks for the info... is it documented in Rev ?
> I don't remember reading it anywhere...

Not per se, or at least nothing connected with the "sort container" 
command that I could find.

Given how big 64k is it rarely comes up in common use; I found out about 
it from a support email from Scott Raney after one of my customers was 
having an inexplicable issue with WebMerge.

It's worth noting that for all customers I have and all the strange 
things they do with WebMerge (I have one customer go generates 300,000 
pages at a time with it) the 64k-per-line limit only came up once in the 
three years I've been selling the product.

In that customer's case we were able to recommend a better workflow for 
them:  they had one field that contained short stories, but it made 
their work more flexible to simply leave the short story in a separate 
file and reference the file from their database, allowing them to edit 
the story file without going into the database.

In recent years I've been using lists as RAM-based database tables with 
good results, and haven't had any record get even close to 64k as long 
as I provide support for referencing external files for things better 
suited for those.

Not all lists will lend themselves to that, but when they do you get to 
keep enjoying the simplicity and efficiency of chunk expressions for 
everything else, and a pretty good sort command. :)

--
  Richard Gaskin
  Managing Editor, revJournal
  _______________________________________________________
  Rev tips, tutorials and more: http://www.revJournal.com



More information about the use-livecode mailing list