Groups | Search | Server Info | Keyboard shortcuts | Login | Register [http] [https] [nntp] [nntps]

Groups > linux.debian.user > #200871 > unrolled thread

ls -la sort order

Back to article view | Back to linux.debian.user

Contents

  ls -la sort order Lee <ler762@gmail.com> - 2018-10-03 17:10 +0200
    Re: ls -la sort order Greg Wooledge <wooledg@eeg.ccf.org> - 2018-10-03 17:20 +0200
      Re: ls -la sort order Lee <ler762@gmail.com> - 2018-10-03 18:40 +0200
        Re: ls -la sort order Greg Wooledge <wooledg@eeg.ccf.org> - 2018-10-03 18:40 +0200
        Re: ls -la sort order Roberto C. Sánchez <roberto@debian.org> - 2018-10-03 18:50 +0200
          Re: ls -la sort order Lee <ler762@gmail.com> - 2018-10-03 20:00 +0200
            Re: ls -la sort order Roberto C. Sánchez <roberto@debian.org> - 2018-10-03 20:10 +0200
              Re: ls -la sort order Lee <ler762@gmail.com> - 2018-10-03 21:00 +0200
            Re: ls -la sort order Greg Wooledge <wooledg@eeg.ccf.org> - 2018-10-03 20:10 +0200

#200871 — ls -la sort order

I really don't like that 'ls -la' seems to ignore a leading dot when
sorting files for display - eg. in terminal I'll get
  old
  Pictures
  .profile
  Public

I haven't changed any settings yet; I've still got LANG=en_US.utf8

I can fix the problem by aliasing ls to 'LC_COLLATE=C  ls' but that
seems klunky and would only fix ls (and not break anything else).  How
bad of an idea would it be to set
  LC_COLLATE=C
in my .bashrc or is there some other setting to get dot files listed
first, then uppercase, then lowercase?


and I thought character codes 32-126 were the same thing in ascii and
utf8, so why is this happening in the first place?

Thanks,
Lee

[toc] | [next] | [standalone]

#200872

On Wed, Oct 03, 2018 at 11:07:04AM -0400, Lee wrote:
> I can fix the problem by aliasing ls to 'LC_COLLATE=C  ls' but that
> seems klunky and would only fix ls (and not break anything else).  How
> bad of an idea would it be to set
>   LC_COLLATE=C
> in my .bashrc or is there some other setting to get dot files listed
> first, then uppercase, then lowercase?

This will affect ls, and also the expansion of globs.  It's totally fine,
assuming this is what you want, and assuming you don't have a lot of
files with non-ASCII characters in their names.  This is exactly what
locale settings are for.

Don't forget to export it.

[toc] | [prev] | [next] | [standalone]

#200873

On 10/3/18, Greg Wooledge <wooledg@eeg.ccf.org> wrote:
> On Wed, Oct 03, 2018 at 11:07:04AM -0400, Lee wrote:
>> I can fix the problem by aliasing ls to 'LC_COLLATE=C  ls' but that
>> seems klunky and would only fix ls (and not break anything else).  How
>> bad of an idea would it be to set
>>   LC_COLLATE=C
>> in my .bashrc or is there some other setting to get dot files listed
>> first, then uppercase, then lowercase?
>
> This will affect ls, and also the expansion of globs.

interesting... I get different results for 'ls [D-M]*' if LC_COLLATE=C
or LC_COLLATE=en_US.utf8

>  It's totally fine,
> assuming this is what you want, and assuming you don't have a lot of
> files with non-ASCII characters in their names.

I'm guessing no since I don't know how to create a filename with a
non-ascii character :)

>  This is exactly what locale settings are for.

cool.. Thank you
Lee

[toc] | [prev] | [next] | [standalone]

#200874

On Wed, Oct 03, 2018 at 12:31:01PM -0400, Lee wrote:
> I'm guessing no since I don't know how to create a filename with a
> non-ascii character :)

This may/may not apply to you, but someone reading the archives may be
interested:

In X, you can type characters that are a superset of what's on your
keyboard using the Compose feature.  This is described (briefly) at
<https://wiki.debian.org/XCompose>.

[toc] | [prev] | [next] | [standalone]

#200875

On Wed, Oct 03, 2018 at 12:31:01PM -0400, Lee wrote:
> 
> interesting... I get different results for 'ls [D-M]*' if LC_COLLATE=C
> or LC_COLLATE=en_US.utf8
> 
Think of it this way:

en_US.utf8 -> sort in alphabetical order
C -> sort in ASCII-betical order

In ASCII, all of the capital letters precede all of the lowercase
letters.  In US English, there is not a sorting distinction between a
capital and its matching lowercase, so they are considered equivalent.
Said another way, in ASCII 'A' != 'a', but en_US.utf8 'A' == 'a' (and
probably in every locale that uses Latin-1 as a base).

Regards,

-Roberto

-- 
Roberto C. Sánchez

[toc] | [prev] | [next] | [standalone]

#200877

On 10/3/18, Roberto C. Sánchez <roberto@debian.org> wrote:
> On Wed, Oct 03, 2018 at 12:31:01PM -0400, Lee wrote:
>>
>> interesting... I get different results for 'ls [D-M]*' if LC_COLLATE=C
>> or LC_COLLATE=en_US.utf8
>>
> Think of it this way:
>
> en_US.utf8 -> sort in alphabetical order
> C -> sort in ASCII-betical order
>
> In ASCII, all of the capital letters precede all of the lowercase
> letters.  In US English, there is not a sorting distinction between a
> capital and its matching lowercase, so they are considered equivalent.
> Said another way, in ASCII 'A' != 'a', but en_US.utf8 'A' == 'a' (and
> probably in every locale that uses Latin-1 as a base).

Sure - I can understand some people wanting A a to sort together.  But
ignoring non-alpha characters when sorting???  Eventually I'm sure I
can get used to
  Music
  old
  Pictures

but this order is obnoxious
  .mozilla
  Music
  old
  Pictures
  .profile
  Public

I don't think I'll ever get used to that.  I'm just a bit concerned
that setting LC_COLLATE=C is going to break something & I'll have a
heck of a time figuring out it was because I changed the sort order.

Regards,
Lee

[toc] | [prev] | [next] | [standalone]

#200878

On Wed, Oct 03, 2018 at 01:51:54PM -0400, Lee wrote:
> 
> Sure - I can understand some people wanting A a to sort together.  But
> ignoring non-alpha characters when sorting???  Eventually I'm sure I
> can get used to
>   Music
>   old
>   Pictures
> 
> but this order is obnoxious
>   .mozilla
>   Music
>   old
>   Pictures
>   .profile
>   Public
> 

While it may not particularly intuitive, I would not call it obnoxious.
In English, the '.' would not customarily be used to make lexical
sorting decisions.

Dr. Jiles
Dr Jones
Dr. Jurgenson

That would be the sort order I expect (for English, or most any other
natural language locale).

Either of

Dr Jones
Dr. Jiles
Dr. Jurgenson

or

Dr. Jiles
Dr. Jurgenson
Dr Jones

would seem incorrect to me.  I suspect that collation function has no
way of knowing whether the punctuation is important or not, so it must
adhere to the rules specified in the locale (to ignore it in this case).
You could always write a custom locale that sorts punctuation before or
after letters, as you like.  The locale(5) man page would be a good
place to start for that.

> I don't think I'll ever get used to that.  I'm just a bit concerned
> that setting LC_COLLATE=C is going to break something & I'll have a
> heck of a time figuring out it was because I changed the sort order.
> 

Call me old school, but I have LC_COLLATE=C in every shell profile on
every machine that I use.  I find that sorting is most important to me
at the shell prompt (and usually when using ls).  There has never been
instance where I encountered an oddity that I even suspected related to
my choice of LC_COLLATE.

Regards,

-Roberto

-- 
Roberto C. Sánchez

[toc] | [prev] | [next] | [standalone]

#200880

On 10/3/18, Roberto C. Sánchez <roberto@debian.org> wrote:
> On Wed, Oct 03, 2018 at 01:51:54PM -0400, Lee wrote:
>>
>> Sure - I can understand some people wanting A a to sort together.  But
>> ignoring non-alpha characters when sorting???  Eventually I'm sure I
>> can get used to
>>   Music
>>   old
>>   Pictures
>>
>> but this order is obnoxious
>>   .mozilla
>>   Music
>>   old
>>   Pictures
>>   .profile
>>   Public
>>
>
> While it may not particularly intuitive, I would not call it obnoxious.
> In English, the '.' would not customarily be used to make lexical
> sorting decisions.
>
> Dr. Jiles
> Dr Jones
> Dr. Jurgenson
>
> That would be the sort order I expect (for English, or most any other
> natural language locale).

Which is probably why I'm having such a hard time accepting the
en_US.utf8 sort order.  I want the "computer" rules, not the natural
language rules.

> Either of
>
> Dr Jones
> Dr. Jiles
> Dr. Jurgenson
>
> or
>
> Dr. Jiles
> Dr. Jurgenson
> Dr Jones
>
> would seem incorrect to me.

<grin> what seems incorrect to me is that the data isn't normalized.
I'd do something like
  sed -e 's/^Dr /Dr. /' | sort

>  I suspect that collation function has no
> way of knowing whether the punctuation is important or not, so it must
> adhere to the rules specified in the locale (to ignore it in this case).
> You could always write a custom locale that sorts punctuation before or
> after letters, as you like.  The locale(5) man page would be a good
> place to start for that.

ok - I'll take a look

>> I don't think I'll ever get used to that.  I'm just a bit concerned
>> that setting LC_COLLATE=C is going to break something & I'll have a
>> heck of a time figuring out it was because I changed the sort order.
>>
>
> Call me old school, but I have LC_COLLATE=C in every shell profile on
> every machine that I use.  I find that sorting is most important to me
> at the shell prompt (and usually when using ls).  There has never been
> instance where I encountered an oddity that I even suspected related to
> my choice of LC_COLLATE.

Good to know!  Thank you
Lee

[toc] | [prev] | [next] | [standalone]

#200879

On Wed, Oct 03, 2018 at 01:51:54PM -0400, Lee wrote:
> I don't think I'll ever get used to that.  I'm just a bit concerned
> that setting LC_COLLATE=C is going to break something & I'll have a
> heck of a time figuring out it was because I changed the sort order.

As a user, it's your prerogative to set your locale variables however
you like.  Setting one of them to C is *always* permitted.

All that matters is what you prefer.

wooledg:~$ printf %s\\n o O oh Ohio ô Ô ó Ó | LC_COLLATE=C sort
O
Ohio
o
oh
Ó
Ô
ó
ô
wooledg:~$ printf %s\\n o O oh Ohio ô Ô ó Ó | LC_COLLATE=en_US.utf8 sort
o
O
ó
Ó
ô
Ô
oh
Ohio

[toc] | [prev] | [standalone]

Back to top | Article view | linux.debian.user

csiph-web