Sorting algorithm needs attention

[foxidrive]

From time to time I've noticed that TC has problems when sorting filenames, so I'm offering an illustration of where I see the issue in TC V6.03

Paste the following into a batch file and execute it in an empty folder. The sort will be in the order that is listed below, where MBYR-ILO.TXT should not follow MB-YADAS.TXT. The Y comes after the - in the ascii table and so the MBYR-ILO.TXT should come after MZ-XXX.TXT
Ditto for MCE-DREA.TXT


@echo off
type nul>MB-THESA.TXT
type nul>MB-WOODY.TXT
type nul>MB-YADAS.TXT
type nul>MBYR-ILO.TXT
type nul>MC-ITSGE.TXT
type nul>MC-MYFOO.TXT
type nul>MC-NACL.TXT
type nul>MC-SOMED.TXT
type nul>MC-WHENE.TXT
type nul>MCE-DREA.TXT
type nul>MD-LILIM.TXT
type nul>MD-SPANI.TXT

[karlchen]

@foxidrive

You are right:
The dash or hyphen has got ASCII value 45, the letter Y has got the ASCII value 89. So "-" is smaller than "Y".

You are wrong with respect to the sorting order nevertheless.
"MBY" comes before "MC-", because sorting is done on character by character base.
And everything starting with "MB" is "smaller" than anything starting with "MC".

So TC sorts the filenames in the correct way.

Kind regards,
Karl

[foxidrive]

You are quite right. I verified it using MS sort.com and TC's output is right.

Thanks for pointing out my error.

[foxidrive]

When gathering newsgroup pictures I noticed this anomoly.
Try this batch file. Horse appears at the beginning when using the MS Sort.

@echo off
type nul>"Animals - Bovine.jpg"
type nul>"Animals- Horse.jpg"
type nul>"Animals - Snake.jpg"

[pdavit]

type nul>"Animals- Horse.jpg"

Maybe because of the missing space?! I don’t know!

[foxidrive]

I'm not sure that I explained myself clearly,
TC V6.03 sorts like this:

Animals - Bovine.jpg
Animals- Horse.jpg
Animals - Snake.jpg

Wheras the sort utility sorts like this:

Animals- Horse.jpg
Animals - Bovine.jpg
Animals - Snake.jpg

If sorting by ascii value the space (32) comes before the hyphen (45), so sort doesn't conform to ascii values and there is wiggle room, but TC shows the hyphen in between two spaces which afaics is wrong. It should either be at the top, to match sorts output, or at the bottom to follow the ascii values.

[SanskritFritz]

And Explorer sorts them like this:

Animals - Bovine.jpg
Animals - Snake.jpg
Animals- Horse.jpg

Hmm

[SanskritFritz]

Do you have SortUpper=2 in your wincmd.ini?

[foxidrive]

D'oh! I did indeed. Have removed it now so I don't make similar blunders in future.

I found this quote from IGL which explains the behaviour.
"Normally the first TC example is standard sorting.
But with SortUpper=2 you want to have different behaviour.
Numbers are sorted in more "humanoidal" way, eg: 1,5,9,10,11,20 intead of standard sorting: 1,10,11,20,5,9.
Also spaces are put first in standard sorting (before letter a), but in SortUpper=2 mode space are ignored"

So it was the spaces being ignored , which puts H after B in Bovine, and before S in Snake, as TC displayed it.
Thanks for the solution.

[SanskritFritz]

I actually very much like the way natural sorting does its job. IMHO the file list is much more readable this way:

file1.txt
file2.txt
file10.txt

[vlado]

One more example:

@echo off
type nul>"a.txt"
type nul>"b.txt"
type nul>"b zzz.txt"
type nul>"c.txt"

Why by abc order "b zzz.txt" stays before b.txt?

[Sheepdog]

Also annoying is:

Amt
horse
postings
ärger

ärger should be sorted like aerger or at least as arger

sheepdog

[Clo]

2Sheepdog
Hello Stefan !
• It's the same with all extended characters, I get:
Ancre21.bmp
Big21.bmp
ESPACE21.BMP
Et-commercial.bmp
O-trema-Maj.bmp
œ.bmp
À.bmp
Where À might be the second, instead the last…
and œ.bmp before O-trema-Maj.bmp, since the dash is ignored with SortUpper=2
* That doesn't seem easy to fix up, I hope it's possible…

 Friendly,
Claude

[vlado]

Also annoying is:

Amt
horse
postings
ärger

ärger should be sorted like aerger or at least as arger

sheepdog

Strange. I have tested it on standard US WinXP and order is as expected:
Amt
ärger
horse
postings

(Same on my slovak localized Win2000.)

Vlado

[Sheepdog]

Also annoying is:

Amt
horse
postings
ärger

ärger should be sorted like aerger or at least as arger

sheepdog

Strange. I have tested it on standard US WinXP and order is as expected:
Amt
ärger
horse
postings

(Same on my slovak localized Win2000.)

Vlado

And you did set SortUpper =2?

I like this sorting except the strange 'Ä' handling.

sheepdog