Understanding Sort Sequence 2 - Code Point

[therube]

Sorting has suddenly sorted my thoughts on sorting. Hello, is anyone out there?
In particular, & particularly, I'm not understanding Everything's sort - can you say, hel-lo?
I think hel-lo has landed me in <hot water>, but I think if I ponder it enough I'll be sorted.

.

Everything - SORT, in what order is sort sorting.png (56.73 KiB) Viewed 1151 times

Understanding Sort Sequence 2
.

Everything - SORT2, d is dictionary, sort is Windows.png (15.13 KiB) Viewed 1151 times

.

Everything - SORT3.png (58.14 KiB) Viewed 1151 times

[void]

-

and

'

are "WEAK" punctuation.
They are ignored when sorting.

This sorting change was made in Everything 1.5.0.1401a to make sorting match Windows Explorer.



Everything sorts by converting single Unicode characters to a weight and sorting by the weight.
- and ' have no weights and are ignored.

If sorting by weight matches, then Everything falls back to a Unicode Code Point sort. (Hel-lo < Hello)



If you want the old Unicode code point sort, please try:

a:=HEX($name:) sort:a

[therube]

Wikipedia: code point, codepoint or code position
What's the difference between a character, a code point, a glyph and a grapheme?
Let's Stop Ascribing Meaning to Code Points

code point, codepoint, codepoints

[therube]

- and ' are "WEAK" punctuation. They are ignored when sorting.

Sorting.
And different from searching.
So a search for: hello, is not going to find hel-lo.
OK.


that's very odd, IMO, & potentially very weird & confusing. on a short list of
file names, you might catch it (i didn't), but on a longer list, i'd think one
is apt to say; where did it go, what happened, where is it ???

as it is, at first i thought it had to do with !PUNC, which i did have enabled,
but then there was no change after unsetting that. & then i thought maybe PATH
was playing in, but that too didn't pan out.

and Windows does that. never knew that. never even heard anyone mention it in
any manner. (But, I'm not anyone, so I will speak out .)

now, i understand the point about code points (actually i don't, but i believe
i understand it on a conceptual basis) & in that respect... but, i still think
it is wrong.


heh, & you mean to tell me that Windows explorer does just that. never would
have known. &, of course, just because they do it, does not make it right.
(maybe there should not be right nor write. simply simplify things to only,
rite, right or did i mean, write .)

(some languages, or at least some writings, at times, do not even use vowels,
so we could simplify even more - rt )


say you have a set of filenames like below
(& where d2f8... ... d2ff... are all the "hex" possibilities, as filenames)

Code: Select all

D02_-_D
d2f85835
...
d2ffffff
D03_-_D

that's a WHOLE lot of d2's before you would ever find the "corresponding" file to D02_-_D, D03_-_D

granted in Windows most everything deals with directories
so if one assumes that D0?_-_D files were in one directory,
& d2* files were in a different directory, you're not apt
to run into this situation, but Everything sees everything

is there even a Windows utility that can search for file
names, globally & display them (sort them) in a Windows
Explorer-like format?


that is real odd, to me, regardless of how "right" it may be
& again, suppose there were some 1000 sH5-* entries

Code: Select all

sh01
sH5-wnbObpnkf4JB
sh006_300_004
sh006_300_007
sh09
sh10
sh11
sh12 - sam1
sh14
sh62

as it was, i FAYT 'sh01' (Fast Find, whatever it is called), which brought sh01 up, centered in my window.
but because of my window size, i only saw this (from the center, down):

Code: Select all

sh01
sh01
sH5-wnbObpnkf4JB
sH5-wnbObpnkf4JB
sh006_300_004
sh006_300_004
sh006_300_007
sh006_300_007

& my first thought was, where is "sh09", which i knew existed, but
happened to be in an overflow area so there was only disconnect,
where i was expecting to see:

Code: Select all

sh01
sh01
sh09
sh09
sh10
sh10

so it was like, huh?

[void]

hello

will match

hel-lo

when Search -> Ignore Punctuation is enabled.



Code Point = The internal value of the character. It has no special meaning.
Weight = Everything sorts every single character using a language neutral comparison. The weight is the index into this sorted table.



Everything will compare a sequence of digits as a whole number.
006 < 09

Ideally, you would want to sort HEX by code point.
However, it's not possible to know if the name is intended to be HEX.
Most filenames will not be in HEX.
When working with HEX filenames, use:

a:=HEX($name:) sort:a

Are you wanting an option to disable natural sort?
I could do an advanced option to disable the natural sort, but you would have to rebuild your index.

[therube]

Are you wanting an option to disable natural sort?

Not particularly.

I could do an advanced option to disable the natural sort, but you would have to rebuild your index.

And in particular, because of that.

I'm just more noting that depending on circumstances one might see very strange looking results.
Or not find expected results, which are in fact there, just not where one is looking.

codepoints, numeric sorts, or whatnot may be good for efficiency, determinabilty <sp>, but not necessarily for a person to comprehend.