Old Problem With Filename Sorting Algorithm

FormalDaHyde · Joined: Jan 02, 2005 Posts: 1511

During the ice ages leading to the WindowsXP OperatingSystem, Neanderthal Windows OS employed a SORTing algorithm that was BINARY in nature.
Fast-forward to WindowsXP: It employs a new Character-based file sorting algorithm, which preserves the logical order of FileNames.
How is that for a factoid??? Wink

Why do I tell you this?
Well, the problem used to become evident when you attempted to SORT FileNames containing numbers.
My standard FileNaming Convention {data that I generate/save} is to always start FileName with a leading 5-digit date code (mmddyy format, unfortunately).

The following is an example of the problem using 3 imaginary files using my FileNaming Convention:
*In older Windows OS, inside WinExplorer (aka FileManager) >> the FileName order sorted/displayed would have been:
010106_readme.txt
100105_readme.txt
020105_readme.txt

*WindowsXP OS, inside the WinExplorer >> the FileName order sorted/displayed would be:
010106_readme.txt
020105_readme.txt
100105_readme.txt

There are some deep settings inside the WinXP registry that allows you to go back to the old method (a user-based setting). But I can't imagine why anyone would want to do that...

I donot know how Linux handles this BINARY versus Character-based dilemma, but I would guess it to be a BINARY sort... ?

Webb · Joined: Feb 17, 2003 Posts: 4271

TweakUI has a setting that will let you disable "intuitive filename sorting".

Something like this, I assume.

But I don't know why anyone would want to do this.

FormalDaHyde · Joined: Jan 02, 2005 Posts: 1511

Xteq X-Setup utility also provides tweaking of the same but it is no longer a FREEware. Thus, I resisted suggesting it! But the cat's out of the bag!

Webb · Joined: Feb 17, 2003 Posts: 4271

I'm here for you. Here is the last freeware version of X-Setup.

Jackie78 · Joined: Dec 18, 2008 Posts: 1

Hi, I think your example is bad, since it wouldn't be sorted the way you describe, neithr the old nor the new way:

010106_readme.txt
100105_readme.txt
020105_readme.txt

This will never happen, even not the old binary way, since 1 is greater than 0 only regarding the first letters, so 100105 will always be last. A better example might be this:

9x.txt
1234x.txt

Sorted Alphapetically, 9 is first, but sorting it numerically 1234x.txt is first. well, this may make sense if you have ONLY numbers, but what about filenames like abx9xyz.txt and abc12xyz.txt? Is it still logical to sort the 9 first? To me, the alphabetical asset makes more sense, but this may be due to the fact that I have grwon up using computers.