sort by length of line (Weird)

[Richard Gaskin]

Mark Wieder wrote:

> I'm sort of OK with it because there's a simple way to get the desired
> answer. But I filed bug 9910 requesting the simplified syntax with
> "numeric" being optional.

Respectfully, I think it may be asking a bit much of the engine - and 
perhaps the scripter - to expect the engine to make that sort of guess 
correctly in all cases.

When we write:

    sort myList by length(each)

...what we're asking the engine to do is two steps:

1. Obtain a list of integers representing the length of each line
2. Sort that list

While both instructions appear in the same statement, they're pretty 
much independent of one another, effectively the same as if they were on 
separate lines:

   get LineLengths(myList) -- some function that works on the list
   sort it

So in effect, request #9910 is asking for the engine to examine the 
values of any list passed to the sort command, and if it finds that 
they're all integers then sort then numerically rather than alphabetically.

On the one hand I agree this would be convenient, since most of the time 
that we do sorts on lists consisting of only integers we're doing so 
numerically.

But on the other hand it's quite a departure from the established sort 
convention in LiveCode, in which it always sorts alphabetically unless 
it's explicitly told to do otherwise.

If this request were implemented as requested, it would introduce a 
"sometimes" rule, in which the engine sometimes sorts alphabetically and 
sometimes not, and the developer would have to keep track of their data 
to reliably know what the engine will choose to do.

Taking the request a step further, we might reasonably expect a list of 
dates to be sorted by dateTime, but once again if this were the default 
behavior we'd have to monitor our data to know what the engine's default 
behavior would be.

At a minimum, it may be useful to have changes in the default sorting 
behavior governed by a global property, something like "the smartsort", 
which would only introduce this "sometimes" rule once it's set, and 
otherwise preserve the requirement for explicit sort typing which has 
characterized the language thus far.

--
  Richard Gaskin
  Fourth World
  LiveCode training and consulting: http://www.fourthworld.com
  Webzine for LiveCode developers: http://www.LiveCodeJournal.com
  LiveCode Journal blog: http://LiveCodejournal.com/blog.irv