bug#52453: Full timezone name

[積丹尼 Dan Jacobson]

>>>>> "PE" == Paul Eggert  writes:
PE> On 12/12/21 07:27, 積丹尼 Dan Jacobson wrote:
>> Also mention something about how to get the full timezone name.

PE> Unfortunately at the coreutils level that information is not available
PE> in any portable or reliable way.

OK. Good. At the bottom of the page do mention that.
Else people wonder why it is not one of the choices.

bug#52453: Full timezone name

[積丹尼 Dan Jacobson]

On (info "(coreutils) Time conversion specifiers")

‘%Z’
 alphabetic time zone abbreviation (e.g., ‘EDT’), or nothing if no
 time zone is determinable.  See ‘%z’ for how it is determined.

Also mention something about how to get the full timezone name.

E.g., show instead of CST: Central Standard Time, or China Standard Time, or
China Central Standard Time, etc.

bug#51288: Break date SYNOPSIS into two sections

[積丹尼 Dan Jacobson]

On the 'date' man and info pages,

SYNOPSIS
   date [OPTION]... [+FORMAT]
   date [-u|--utc|--universal] [MMDDhhmm[[CC]YY][.ss]]

Synopses:

 date [OPTION]... [+FORMAT]
 date [-u|--utc|--universal] [ MMDDhhmm[[CC]YY][.ss] ]

please break these down into:

printing the date:

 date [OPTION]... [+FORMAT]

setting the date:

 date [-u|--utc|--universal] [ MMDDhhmm[[CC]YY][.ss] ]
 date [-s|--set] DATESTR

else the user in a panic to set the date, will try
# date -s 10140110
date: cannot set date: Invalid argument
Mon Jan 10 00:00:00 LMT 1014
Instead of
$ date10140110

Only on
(info "(coreutils) Setting the time")
is the mystery revealed. P.S., on that page you should mention the
MMDDhhmm... string again, not just its components.

bug#50694: ls and cpio's idea of "six months ago" are slightly different

[積丹尼 Dan Jacobson]

What a headache.
"Six months ago" means slightly different things to cpio and ls.
And ls documents do say exactly what,
and cpio documents don't even say six months.

$ cat prover
set -eu
cd /tmp
for i in `seq 170 190`
do
touch -d "$i days ago" x
{
env - ls -l x|tr -s ' '
ls x|cpio -o 2>&-|cpio -tv 2>&-|tr -s ' '
} > z
if test `sort -u z|wc -l` != 1
then cat z; echo
fi
done
$ sh prover
-rw-r--r-- 1 jidanni jidanni 0 Mar 23 12:47 x
-rw-r--r-- 1 jidanni jidanni 0 Mar 23 2021 x

-rw-r--r-- 1 jidanni jidanni 0 Mar 22 12:47 x
-rw-r--r-- 1 jidanni jidanni 0 Mar 22 2021 x

So two things that should look the same ... don't.

What's worse is there is no option to change cpio's date format.

Well then just change ls's date format to match cpio's right?
Well yes, but then there is no way to change ls's idea of six months,
even if we examined cpio's source code to find out what it uses.

$ ls --version
ls (GNU coreutils) 8.32
Copyright (C) 2020 Free Software Foundation, Inc.

$ cpio --version
cpio (GNU cpio) 2.13
Copyright (C) 2017 Free Software Foundation, Inc.

bug#49764: tr invocation INFO page missing list of options

[積丹尼 Dan Jacobson]

(info "(coreutils) tr invocation")
all the options don't appear together on the same page.
Only the man page's DESCRIPTION has them all together.

bug#49503: Mention workarounds, so one could achieve the Debian version sorting algorithm

[積丹尼 Dan Jacobson]

(info "(coreutils) Differences from the official Debian Algorithm") and
(info "(coreutils) Minus/Hyphen and Colon characters")
could mention workarounds, so one could indeed achieve the Debian
Algorithm.
Or mention the only way is to use
dpkg --compare-versions
(on pairs only.)

bug#48248: tr docs: mention what to expect now vs. future

[積丹尼 Dan Jacobson]

(info "(coreutils) tr invocation") says
   Currently ‘tr’ fully supports only single-byte characters.
   Eventually it will support multibyte characters; when it does, the ‘-C’

And the man page has no warning at all. #BUG. Please add one.

BTW,

$ echo 哇|tr 哇 123
123

It works fine! (People will think.)

Therefore be sure to mention just when it will start failing, with real 
examples,

$ echo 哇|tr 哇 1
111
$ echo 哇|tr 哇 1234
123
$ echo 哇哇|tr 哇 1234
123123
$ echo 哇哇姑五湖|tr 哇 1234
1231231__五湖

$ echo "22°56'52.2\"N 121°14'09.3\"E" | tr ° d
22dd56'52.2"N 121dd14'09.3"E

Also on the INFO page, give examples of, for the same input,
what to expect now, and what to expect in the future.

bug#47891: sort --numeric-sort-Extra-Strength

[積丹尼 Dan Jacobson]

Let's face it, sort, no matter what --option, or LC_... value,
just can't achieve this order:

3－1號邊
3號之1
3號之2
30

What's the miracle kid (me) doing right that sort(1) just can't sort
out yet?

Mom told me "Just sort by numbers."

See https://www.jidanni.org/geo/house_numbering/sorting/ in your spare time.

bug#47859: Additional seq outlandish example: seq 0 dangers

[積丹尼 Dan Jacobson]

On (info "(coreutils) seq invocation") we read
   Be careful when using ‘seq’ with outlandish values: otherwise you
   may...

Here's another 'fun/sad/DDOS yourself' example you might add:

One day I wrote a Makefile,
m:
seq 0 9|sed s/$$/號.html/|xargs make
but before using it, I though I'll just test with one item,
m:
seq 0  |sed s/$$/號.html/|xargs make
well of course... as seq prints nothing here,
this triggered a massive ever growing recursive loop...

Yes, all my fault for picking 0. I'll pick 1 next time.

P.S., perhaps document how to get seq to cough up just "0". One way I
found was:
$ seq 0 1 0
0

bug#47858: Document that FULLWIDTH DIGITs boggle coreutils brain currently

[積丹尼 Dan Jacobson]

(info "(coreutils) Correct/Incorrect ordering and Expected/Unexpected results")
needs to "admit" or "at least mention" that "coreutils just hasn't thought 
about FULLWIDTH
DIGITs yet." I.e., Europeans are in luck. East Asians are out of luck.

$ seq 9 11|sort
10
11
9
$ seq 9 11|sort -n
9
10
11
$ seq 9 11|sort -V
9
10
11

$ echo -e ９\\n１０\\n１１|sort
１０
１１
９
$ echo -e ９\\n１０\\n１１|sort -V
１０
１１
９
$ echo -e ９\\n１０\\n１１|sort -n
１０
１１
９

$ unicode ９
U+FF19 FULLWIDTH DIGIT NINE
UTF-8: ef bc 99 UTF-16BE: ff19 Decimal:  Octal: \0177431
９
Category: Nd (Number, Decimal Digit); East Asian width: F (fullwidth)
Unicode block: FF00..FFEF; Halfwidth and Fullwidth Forms
Numeric value: 9
Digit value: 9
Bidi: EN (European Number)

Decomposition:  0039

Nope, LC_ALL=zh_TW.UTF-8 didn't help.
sort (GNU coreutils) 8.32

bug#47700: Backup options vs. file extensions

[積丹尼 Dan Jacobson]

(info "(coreutils) Backup options") is great, except,
it needs examples of how one can preserve file extension with it.

E.g., x.jpg~ messes up plenty of viewers, but x~.jpg is fine.
Same with numbered backups.

So the page should discuss this problem, and mention workarounds.

bug#47703: Mention if ln -s needs -r, or ln -r needs -s

[積丹尼 Dan Jacobson]

man ln says:

   -r, --relative
  create symbolic links relative to link location

   -s, --symbolic
  make symbolic links instead of hard links

Does -s need -r,
or   -r need -s?

Hard to tell from the man page.

Had to find out the 'hard' way.

Please mention the answer on the man page.

Thanks.

bug#47702: wc man page: first you are talking about bytes, then you are talking about characters

[積丹尼 Dan Jacobson]

Man wc says

   Print newline, word, and byte counts for each FILE, and a total line if
   more than one FILE is specified.  A word is a non-zero-length  sequence
   of characters delimited by white space.

first you are talking about bytes, then you are talking about
characters.

So for the latter, please say
characters (not bytes)
or
characters (same as bytes)
or just
bytes
Yes, even if explained in the INFO file.
Thanks.

bug#47701: Add ln --dry-run

[積丹尼 Dan Jacobson]

$ ln -rbsv ~/Downloads/*/95*[0-9].jpg xiangpianjibentu
'xiangpianjibentu/95212025.jpg' -> 
'../../../../Downloads/ViewLarge_Draw_files/95212025.jpg'
'xiangpianjibentu/95212033.jpg' -> 
'../../../../Downloads/ViecccwLarge_Draw_files/95212033.jpg'
'xiangpianjibentu/95212025.jpg~' ~ 'xiangpianjibentu/95212025.jpg' -> 
'../../../../Downloads/ViewLarge_Draw_files/95212025.jpg'
'xiangpianjibentu/95212026.jpg' -> 
'../../../../Downloads/ViewLargeddd_Draw_files/95212026.jpg'
'xiangpianjibentu/95212034.jpg' -> 
'../../../../Downloads/ViewLargvvve_Draw_files/95212034.jpg'
'xiangpianjibentu/95212045.jpg' -> 
'../../../../Downloads/ViewxxxLarge_Draw_files/95212045.jpg'

Wow, that was a pretty fancy result.

However the only way to find out what will happen is to cross one's
fingers and run the command and hope for the best.

So perhaps add a --dry-run.

It would print (when -v is used), but not do, the above.

(Except probably for the one backup file.)

Anyway, we all know that sometimes ln syntax is tricky,
so perhaps add --dry-run.

bug#47353: Numbered backups also need kept-new-versions else will grow out of control

[積丹尼 Dan Jacobson]

All I am saying is 'to acknowledge the hazards'... 'Advisory on
prolonged use'. 'Note about health effects of unattended use.'

For all the user knows reading
(info "(coreutils) Backup options")
maybe numbered backups even just means one backup, numbered ...1.

bug#47353: Numbered backups also need kept-new-versions else will grow out of control

[積丹尼 Dan Jacobson]

> "CE" == Chris Elvidge  writes:

CE> So 'use common sense to limit disk space used' has to be stated in the
CE> manual?

Hmmm, yes.
And also mention in the manual that e.g., emacs has methods to trim
these automatically, but coreutils hasn't implemented them yet.

Else users will write elaborate programs to do it, thinking that nobody
thought about it yet.

bug#46346: wc --human-readable or --verbose

[積丹尼 Dan Jacobson]

wc needs a --verbose option. Else one is forced to do:

$ file=e.html; echo $file:; for i in bytes chars lines words; do echo -en 
$i:\\t; wc --$i < $file; done
e.html:
bytes:  31655
chars:  29141
lines:  643
words:  1275

I mean sometimes we want to send the output to a real person, and
currently all wc makes is:

$ wc e.html
  643  1275 31655 e.html
$ file=e.html; for i in bytes chars lines words; do wc --$i < $file; done
31655
29141
643
1275

So there needs to be a --verbose or --human-readable like du(1)... oops
I don't mean saying 234M... yet.

bug#46060: Offer ls --limit=...

[積丹尼 Dan Jacobson]

> "PE" == Paul Eggert  writes:
PE> That argument would apply to any program, no? "cat", "diff", "sh",
PE> "node",

PE> Not sure why "ls" needs a convenience flag that would complicate the
PE> documentation and maintenance and be so rarely useful.

OK, then I'll close the bug then.

bug#46060: Offer ls --limit=...

[積丹尼 Dan Jacobson]

E.g.,
"What is API pagination? Some APIs, such as Contacts can return millions
of results. We obviously can't return all of them at once, so we need to
return a subset - or a page - at a time. This technique is called paging
and is common to most APIs. Paging can be implemented in many different
ways, some better than others."
Anyway, this ls command was built in the early years of computer science...

bug#46060: Offer ls --limit=...

[積丹尼 Dan Jacobson]

Sure, it is against the https://en.wikipedia.org/wiki/Unix_philosophy, but
just like SQL has LIMIT,
and
$ unicode --help
  -m MAXCOUNT, --max=MAXCOUNT
Maximal number of codepoints to display...

Just like "we want to stop pollution at the source", not always "clean up after 
it".

bug#46060: Offer ls --limit=...

[積丹尼 Dan Jacobson]

I hereby propose "ls --limit=..."

$ ls --limit=1 # Would only print one result item:
A

You might say:
"Jacobson, just use "ls|sed q". Closed: Worksforme."

Ah, but I am talking about items, not lines:

$ ls
A  B  C  D
E  F  G  H
I  J  K  L
$ ls -C|sed 2q
A  B  C  D
E  F  G  H
$ ls -C --limit=2
AB
$ ls--limit=2
AB

Indeed, directories might be huge. And any database command already has
a --limit option these days, and does not rely on a second program to
trim its output because it can't control itself. Indeed, on some remote
connections one would only want to launch one program, not two. Thanks.

bug#44975: Warn that fold will wreck UTF-8

[積丹尼 Dan Jacobson]

On (info "(coreutils) fold invocation") warn that fold will wreck UTF-8
files, shattering any multibyte character that unfortunately lies near
its chopper.

(GNU coreutils) 8.32

bug#44959: date error message should say -I

[積丹尼 Dan Jacobson]

Well OK, but when and when not to use the "=" is not revealed by the
otherwise detailed error messages. So unless the user checks the manual,
they will never "get it".

bug#44960: date --expose-flags

[積丹尼 Dan Jacobson]

date +%Y-%m-%dT%H:%M:%S%z
^I meant %:z

> Then one needn't use trial and error

(Driving my point home.)

bug#44960: date --expose-flags

[積丹尼 Dan Jacobson]

I got this brilliant idea.
Let's say one likes the output of

$ date --iso-8601=seconds
2020-11-30T21:15:47+08:00

but wants to know "how you did it?"

Hmmm, no assistance from
$ date --iso-8601=seconds --debug
2020-11-30T21:15:50+08:00

Ah, if only there were a:
$ date --iso-8601=seconds --expose-flags
date +%Y-%m-%dT%H:%M:%S%z

Then one needn't use trial and error to figure out the template being
used behind the scenes!

date (GNU coreutils) 8.32

bug#44959: date error message should say -I

[積丹尼 Dan Jacobson]

$ date -I=seconds
date: invalid argument ‘=seconds’ for ‘--iso-8601’

Hey, that is a valid argument for --iso-8601. (But not for -I, so say
that instead.)

Here is a real invalid argument:

$ date --iso-8601=secondsz
date: invalid argument ‘secondsz’ for ‘--iso-8601’

date (GNU coreutils) 8.32

bug#44770: chown: warn about the dot when encountering it

[積丹尼 Dan Jacobson]

Maybe print warning messages when encountering the dot,
(info "(coreutils) chown invocation")
Else Grandpa won't ever know,
https://github.com/scop/bash-completion/issues/468
until one day when it's too late...
(And his program starts messing things up on some other system.)

bug#40943: nl: add way to exclude matching lines from getting numbered!

[積丹尼 Dan Jacobson]

(info "(coreutils) nl invocation") says
 ‘pBRE’
  number only lines that contain a match for the basic regular
  expression BRE.  *Note Regular Expressions: (grep)Regular
  Expressions.

OK, but add a new functionality:
 ‘PBRE’
  Do NOT number lines that contain a match for the basic regular
  expression BRE.  *Note Regular Expressions: (grep)Regular
  Expressions.

$ nl --version
nl (GNU coreutils) 8.30

bug#40942: Clean up mailing list confusion on web page

[積丹尼 Dan Jacobson]

https://www.gnu.org/software/coreutils/ mentions coreut...@gnu.org and
bug-coreutils@gnu.org (and the term mailing list and the term bug tracker
many times, all intertwined and perhaps repeating.)

Maybe only mention each one once, with a description of what they are
below it.

So that would be a total of two items.

bug#40633: Document / implement sorting via absolute value

[積丹尼 Dan Jacobson]

(info "(coreutils) sort invocation")
should mention how to sort via absolute value,
or add an option to do it.
Currently one must use e.g.,
$ seq -2 2|perl -pe 's/^(-)(.*)/$2$1/;' | sort -n
0
1
1-
2
2-

bug#40363: date: add Julian day number output format

[積丹尼 Dan Jacobson]

date(1) should add "Julian day number" or the "Astronomical day
number" (see (info "(emacs) Calendar Systems") output format.

Else for crontab(5) one must do
   # Run once per 48 hours (every other day), even across week, month, and 
year boundaries:
   33 22 * * * expr $(date +\%s) / 60 / 60 / 24 \% 2 > /dev/null && 
echo Check the toilet tissue.

Instead of just
   33 22 * * * expr $(date +\%J) \% 2 > /dev/null && echo Check the 
toilet tissue.
(assuming %J is the Julian day number.)

Currently you have day of the week, day of the month, day of the year,
seconds from second #1. Years from year #1, but no days from "day #1".

bug#39613: stat(1) shouldn't call permissions "Access..."

[積丹尼 Dan Jacobson]

(info "(coreutils) stat invocation") says,
   • %a - Access rights in octal (note ‘#’ and ‘0’ printf flags)
   • %A - Access rights in human readable form

But man find says,
   -perm mode
  File's permission bits are exactly mode (octal or symbolic).  
Since an exact match is required,  if
  you want to use this form for symbolic modes, you may have to 
specify a rather complex mode string.
  For example `-perm g=w' will only match files which have mode 
0020 (that is, ones for  which  group
  write  permission is the only permission set).  It is more likely 
that you will want to use the `/'
  or `-' forms, for example `-perm -g=w', which matches any file 
with group  write  permission.   See
  the EXAMPLES section for some illustrative examples.

   -perm -mode
  All  of  the  permission bits mode are set for the file.  
Symbolic modes are accepted in this form,
  and this is usually the way in which you would want to use them.  
You must specify `u', `g' or  `o'
  if you use a symbolic mode.  See the EXAMPLES section for some 
illustrative examples.

   -perm /mode
  Any  of  the  permission bits mode are set for the file.  
Symbolic modes are accepted in this form.
  You must specify `u', `g' or `o' if you use a symbolic mode.  See 
the EXAMPLES section for some il‐
  lustrative  examples.   If no permission bits in mode are set, 
this test matches any file (the idea
  here is to be consistent with the behaviour of -perm -000).

   -perm +mode
  This is no longer supported (and has been deprecated since 2005). 
 Use -perm /mode instead.


So stat's man and Info pages shouldn't use "Access" to talk about them.
Especially as it pretty much means something else:

   -readable
  Matches files which are readable by the current user.  This takes 
into account access control lists
  and other permissions artefacts which the -perm test ignores.  
This test makes use of the access(2)
  system call, and so can be fooled by NFS servers which do UID 
mapping  (or  root-squashing),  since
  many  systems  implement access(2) in the client's kernel and so 
cannot make use of the UID mapping
  information held on the server.

bug#39017: Mention more about uniq failure exit status

[積丹尼 Dan Jacobson]

(info "(coreutils) uniq invocation") says
   An exit status of zero indicates success, and a nonzero value
   indicates failure.

The user wonders:
"Failure to make a file unique because it was already unique?"
Be more explicit.

bug#38780: date: next feb 11

[積丹尼 Dan Jacobson]

date -d should add a way to say "next feb 11" or "next 2/11" etc.
E.g., at(1) knows we are always talking about the coming Feb. 11,
and not the one that has already passed in the current year.
Just like we can say "next Friday" and "last Friday" currently.

bug#36220: ls -l: maddening mixed left right justifications with numeric ids

[積丹尼 Dan Jacobson]

> "PE" == Paul Eggert  writes:
PE> On 6/17/19 8:12 AM, Pádraig Brady wrote:

PE> I prefer the current ("maddening") behavior, as it gives the reader a
PE> useful signal that the user is numeric rather than textual. This is
PE> particularly important when a user name consists entirely of digits,
PE> which is allowed on some systems. In that case, left alignment of a
PE> string like "" means it's a textual user name, whereas right
PE> alignment means it's a numeric user ID. Admittedly the disambiguation
PE> is not perfect (as there is no cue when the user name has the maximum
PE> width) but it's helpful in common practice.

PE> I thought this was all documented somewhere - at least, it was a
PE> conscious decision when I wrote that code long ago.

Indeed all you need to do would me mention the logic there on INFO "ls"
and we would realize that this was actually a feature.

bug#36220: ls -l: maddening mixed left right justifications with numeric ids

[積丹尼 Dan Jacobson]

> "PB" == Pádraig Brady  writes:
PB> to know if it was a common issue, as I've not encountered it at least.

Well as the anbox Android emulator could possibly become very popular,
ls should be ready for a big influx of numeric ids.

bug#35748: Add cut FIELD-LIST examples

[積丹尼 Dan Jacobson]

> "EB" == Eric Blake  writes:
EB> No need to open a second bug to tell us that the first was not a problem
EB> after all. I'm merging the two bugs in the database, and marking the
EB> issue closed.

It takes several minutes before the bug number is created, so there is a
tradeoff between proper vs. speedy followups.

bug#35748: Add cut FIELD-LIST examples

[積丹尼 Dan Jacobson]

Oh, I see, it is mentioned above it.
Sure wish "(see above)" was there.

bug#35747: Add cut FIELD-LIST examples

[積丹尼 Dan Jacobson]

On (info "(coreutils) cut invocation")

with no examples for

   ‘-f FIELD-LIST’
   ‘--fields=FIELD-LIST’
Select for printing only the fields listed in FIELD-LIST.  Fields
are separated by a TAB character by default.  Also print any line
that contains no delimiter character, unless the ‘--only-delimited’
(‘-s’) option is specified.

one might even one is to put tabs in our FIELD-LIST, or maybe dashes or
commas. And maybe ranges are not supported, etc.

bug#35343: sort: printing characters: define

[積丹尼 Dan Jacobson]

Well all I know is e.g., perl's man perlrecharclass is
more detailed than
(info "(coreutils) Character sets") section,
(which (info "(coreutils) sort invocation") perhaps should also link to.)
Anyways, coreutils should at least link to a definite statement...

bug#35431: Mention "cp -l a b is exactly equivalent to ln a b"

[積丹尼 Dan Jacobson]

On (info "(coreutils) cp invocation") at

   ‘-l’
   ‘--link’
Make hard links instead of copies of non-directories.

Mention

"Exactly equivalent to ln."

or

 "cp -l a b is exactly equivalent to ln a b."

Or if not, mention why.

bug#35343: sort: printing characters: define

[積丹尼 Dan Jacobson]

(info "(coreutils) sort invocation") says

‘-i’
‘--ignore-nonprinting’
 Ignore nonprinting characters.  The ‘LC_CTYPE’ locale determines
 character types...

Well this leaves the user high and dry trying to figure out what you
mean by printing characters.

Is SPC a printing character he wonders. It makes the printer head
advance one unit, whereas ESC doesn't.

So at least mention the ASCII range [A-Z...] of what your are talking
about. It would only take up a word or two of the INFO page.

Anyway, each program might have their own idea of "printing characters"
as well as grandpa down the hall's definition. So the user is left unsure.

bug#35336: date should respect TIME_STYLE too

[積丹尼 Dan Jacobson]

All the other utilities respect the environment variable ‘TIME_STYLE’,
except date(1) itself! Date should be no different...

bug#35335: touch --verbose

[積丹尼 Dan Jacobson]

Say, rm, cp, mv have --verbose.
How about for touch too?
Then one wouldn't need
$ set -x; touch /tmp/{a,b,c}.el; set +x
+ touch /tmp/a.el /tmp/b.el /tmp/c.el
to see what it was touching or not.

bug#34988: mv: check before asking users useless questions

[積丹尼 Dan Jacobson]

> "LSS" == Leslie S Satenstein  writes:
LSS> Did you consider that more than one terminal can be open, and at the
LSS> second terminal, "b" could be renamed before replying to the mv
LSS> command?

I don't know about all that. Let's hope the designers consider all aspects.

bug#34664: Mention wc default action on man page

[積丹尼 Dan Jacobson]

Well
> Print  newline,  word,  and  byte  counts
might sound like
> Play movies, audio files, and even ...

Sort of like a list of capabilities.

> "PB" == Pádraig Brady  writes:
PB> I'm confused. It says 'bytes', and means it.
PB> One has to specify -m to select characters.

But OK, fine.

bug#34670: Implement "seq infinity"

[積丹尼 Dan Jacobson]

I can do
$ sleep infinity
so why not
$ seq infinity
also?

bug#34669: Document sleep infinity

[積丹尼 Dan Jacobson]

$ man sleep | grep -ic infinity
0
$ sleep --help | grep -ic infinity
0
$ info sleep | grep -ic infinity
0
$ /bin/sleep infinity #But it still works, so please document it.

bug#34664: Mention wc default action on man page

[積丹尼 Dan Jacobson]

INFO says
   By default, ‘wc’ prints three counts: the newline, words, and byte
and wc --help even more so.

Alas man wc doesn't say if what we are looking at is bytes or
characters, so kindly mention the default on the man page!

bug#34488: Add sort --limit, or document workarounds for sort|head error messages

[積丹尼 Dan Jacobson]

Thanks for working on this guys, hopefully users will one day no longer
be faced with (yes, seemingly) random errors.
$ ls -l | sed 2!d
drwxr-xr-x 14 jidanni jidanni  4096 2016-12-24  AndroidMisc
$ ls -l | sed 2q
total 157780
drwxr-xr-x 14 jidanni jidanni  4096 2016-12-24  AndroidMisc
ls: write error: Broken pipe

bug#34490: console the user that his -n in sort --debug -n was seen

[積丹尼 Dan Jacobson]

> "PB" == Pádraig Brady  writes:
PB> Fair point. I'm thinking of this extra qualification:

PB>   sort: text ordering performed using ‘en_IE.UTF-8’ sorting rules
Maybe say  'LC_CTYPE=en_IE.UTF-8' 
sorting rules

PB>   sort: text ordering performed using simple byte comparison

sort: text ordering performed using simple byte (LC_CTYPE=C) comparison
or something like that.

bug#34488: Add sort --limit, or document workarounds for sort|head error messages

[積丹尼 Dan Jacobson]

(I recall I heard about 50 years ago when pipe buffers first came to
Unix they were supposed to be invisible to the user...)

bug#34475: Mention even more worries for test -a

[積丹尼 Dan Jacobson]

By the way
https://lists.gnu.org/archive/html/bug-bash/2019-02/msg00052.html
says "... contradictory description in the coreutils info document."

bug#34488: Add sort --limit, or document workarounds for sort|head error messages

[積丹尼 Dan Jacobson]

> "EB" == Eric Blake  writes:

>> And no fair saying "just save the output" (could be big) "into a file
>> first, and do head(1) or sed(1) on that."

EB> If you have an app that exits noisily on write failures to an early-exit
EB> pipe, your solutions are to quit ignoring SIGPIPE, or to change the
EB> pipeline to consume all of the output instead of exiting early and
EB> causing write failure.

OK, it seems that indeed it would then be best to document:

 If you want to make sure no such errors occur on large output,
 instead of
 $ sort | head
 use
 $ sort > file; head file

as an _easy to remember_ alternative amongst the many.

And OK put this in some general place, and have the sort, etc. pages
link to it. Thanks.

bug#34490: console the user that his -n in sort --debug -n was seen

[積丹尼 Dan Jacobson]

Can you please have assure the user in that message it makes
that it has indeed seen his -n/--numeric-sort.

$ sort --debug
sort: using simple byte comparison
$ sort --debug --numeric-sort
sort: using simple byte comparison
    unchanged. User gets nervous.

sort (GNU coreutils) 8.30. Sorry we discussed this before but kinda hard to 
find...

bug#34488: Add sort --limit, or document workarounds for sort|head error messages

[積丹尼 Dan Jacobson]

> "AG" == Assaf Gordon  writes:
AG> The errors are not "random" - they happen because you explicitly
AG> cut short the output of a program.

Well the user thinks "hey I cut short 5 lines, 45 lines, 495 lines, and
then one I got a job at a big company and cut short 4995 lines and got
this error message just when the boss was looking."

So be sure to mention it. Thanks.

bug#34488: Add sort --limit, or document workarounds for sort|head error messages

[積丹尼 Dan Jacobson]

Things start out cheery, but quickly get ugly,

$ for i in 9 99 999  9; do seq $i|sort -n|sed 5q|wc -l; done
5
5
5
5
sort: write failed: 'standard output': Broken pipe
sort: write error
5
sort: write failed: 'standard output': Broken pipe
sort: write error

Therefore, kindly add a sort --limit=n,
and/or on (info "(coreutils) sort invocation")
admit the problem, and give some workarounds, lest
our scripts occasionally spew error messages seemingly randomly,
just when the boss is looking.

And no fair saying "just save the output" (could be big) "into a file
first, and do head(1) or sed(1) on that."

bug#34475: Mention even more worries for test -a

[積丹尼 Dan Jacobson]

First, on the test(1) man page, at


   EXPRESSION1 -a EXPRESSION2
  both EXPRESSION1 and EXPRESSION2 are true

   EXPRESSION1 -o EXPRESSION2
  either EXPRESSION1 or EXPRESSION2 is true


Say instead

   EXPRESSION1 -a EXPRESSION2
  both EXPRESSION1 and EXPRESSION2 are true (See warning NOTE 
below.)

   EXPRESSION1 -o EXPRESSION2
  either EXPRESSION1 or EXPRESSION2 is true (See warning NOTE 
below.)

else users who never make it down to the "fine print at bottom" will
blithefully use them a lifetime without ever an inkling of the dangers!

(Indeed on (info "(coreutils) Connectives for test")
you discuss the dangers first, before even letting the users eyeballs
touch the syntax!)

Also on both man and info pages you need to warn of portability worries,
e.g., in bash we see -a and -o do even more things.

$ help test|egrep -- ' -(a|o) '
  -a FILETrue if file exists.
  -o OPTION  True if the shell option OPTION is enabled.
  EXPR1 -a EXPR2 True if both expr1 AND expr2 are true.
  EXPR1 -o EXPR2 True if either expr1 OR expr2 is true.

bug#34347: At least add 2nd comm error message after output

[積丹尼 Dan Jacobson]

(info "(coreutils) comm invocation")

"Before ‘comm’ can be used, the input files must be sorted"
 
"If an input file is diagnosed as being unsorted, the ‘comm’ command
will exit with a nonzero status (and the output should not be used).
 ^

"Forcing ‘comm’ to process wrongly sorted input files ... is not
guaranteed to produce any particular output. The output will probably
not correspond with whatever you hoped it would be."

OK, so I would change the default.

The new default should be: before producing even a line of output,
check both files. If either of them is in order, print the error
message, and exit non-zero... _without any line of output!_

Because currently,

$ seq 55 > 5
$ seq 55 |tac > 5r
$ comm 5 5r
just causes the error messages to fly off the screen, leaving only the
savvy user, who checks exit values, to have any inkling that there is a
problem, and he should not order futher warships, airplanes, etc. based
on the results after all.

OR, be so kind as to add a second error message after all the output is
printed, so it won't fly off the screen.

bug#34026: mention that long options aren't always as good as short options

[積丹尼 Dan Jacobson]

Yes do warn in the manual, as here root (so no $HOME) expected tilde
expansion... Thanks.

bug#34026: mention that long options aren't always as good as short options

[積丹尼 Dan Jacobson]

On (info "(coreutils) Common options")
mention that, though clearer, long options aren't always as good as short 
options,
$ touch --reference=~jidanni/.pcmanx /cf/pcmanx_time
touch: failed to get attributes of '~jidanni/.pcmanx/pcmanx': No such file or 
directory
Unless one removes the '=':
$ touch --reference ~jidanni/.pcmanx/pcmanx /cf/pcmanx_time
$
(due to the shell helping out with "~...")

bug#34009: warn that mkdir --mode doesn't affect parents created

[積丹尼 Dan Jacobson]

On man mkdir

   -m, --mode=MODE
  set file mode (as in chmod), not a=rwx - umask

   -p, --parents
  no error if existing, make parent directories as needed

do warn that --mode doesn't affect any parents created.

$ mkdir --mode 700 -p /tmp/g/h/i
$ find /tmp/g -ls
55795  0 drwxr-xr-x   3 jidanni  jidanni60 Jan  7 23:30 /tmp/g
55796  0 drwxr-xr-x   3 jidanni  jidanni60 Jan  7 23:30 /tmp/g/h
55797  0 drwx--   2 jidanni  jidanni40 Jan  7 23:30 
/tmp/g/h/i

Also warn on (info "(coreutils) mkdir invocation") more directly. Thanks.

bug#26902: good invalid dates vs. bad invalid dates

[積丹尼 Dan Jacobson]

Maybe it should try to pass on to the user at least an error code:
Invalid date: error code 0754,
or even better, an mention of what error it hit.

bug#33025: Add examples of why one would want to "sort" something "randomly"

[積丹尼 Dan Jacobson]

> "AG" == Assaf Gordon  writes:

AG> (It doesn't deal with "why", that is left to the users to decide when
AG> they need it, but it shows clear examples of how to use it).

Maybe add "For example, you might want to hear your songs in a fresh fun
order. Or make sure your program still works even if the moms come
between or after the pops..."
https://www.google.com/search?q=reasons+for+random+order

bug#33577: ls lacks null terminator option

[積丹尼 Dan Jacobson]

Fine. Put a message on top of (info "(coreutils) ls invocation")
saying that your pipes are better.

bug#33577: ls lacks null terminator option

[積丹尼 Dan Jacobson]

Are you saying you want to enhance find(1) to sort its output?

bug#33577: ls lacks null terminator option

[積丹尼 Dan Jacobson]

Don't let that ls colors stuff fool you. It originally didn't have that.

bug#33577: ls lacks null terminator option

[積丹尼 Dan Jacobson]

(Probably The Unix Programming Environment by K & P.)

bug#33577: ls lacks null terminator option

[積丹尼 Dan Jacobson]

Bob, I remember from my K Unix book that ls can be used in scripts.
In fact that is why
$ ls
acts different than
$ ls | cat

Plus there are lots of things ls can do that find cannot.

For instance sorting.

How are you going to do
$ ls -t
with find?

bug#33577: ls lacks null terminator option

[積丹尼 Dan Jacobson]

For files with blanks in their names,
one shouldn't need this workaround:
$ ls -t | tr \\n \\0 | xargs -0 more > /tmp/z.txt
Please add a --print0 option. like find(1) has.

ls (GNU coreutils) 8.30

bug#23302: mention what are nonprinting characters

[積丹尼 Dan Jacobson]

Good! You need to then tie all the documentation you found, into the
coreutils documentation, as the official declaration of what you mean.
Just like "man perlrecharclass - Perl Regular Expression Character
Classes" does. I mean one cannot just hope the user will "Google" and
then land on "Wikipedia" and hope what is there is 100% the same.

bug#23302: mention what are nonprinting characters

[積丹尼 Dan Jacobson]

Yes but every program has slightly different sets of non-printing
characters, so they need to list them exactly.

bug#21349: acknowledged by developer (Re: bug#21349: who shows no users nowadays on Debian)

[積丹尼 Dan Jacobson]

(Today on Debian I see
$ who
jidanni  pts/02018-10-25 06:26 (:0)
So maybe this is fixed.)

bug#18479: acknowledged by developer ()

[積丹尼 Dan Jacobson]

As you can see, the Subject got damaged somehow when sending this
message to me.

(Subject: Re: bug#18479 acknowledged by developer ())

> "GbTS" == GNU bug Tracking System  writes:
GbTS> This is an automatic notification regarding your bug report
GbTS> #18479: date -d 'today - month',
GbTS> which was filed against the coreutils package.

GbTS> Thank you for your report, which has now been closed.
GbTS> You can view the full report at
GbTS> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=18479

GbTS> If you require further information, please followup to 
18...@debbugs.gnu.org.

GbTS> debbugs.gnu.org maintainers
GbTS> (administrator, GNU bugs database)

bug#14971: acknowledged by developer (Re: bug#14971: split man page table mushed)

[積丹尼 Dan Jacobson]

Say, can we get the final information just from this email next time,
just like in Debian. Where one doesn't need to click on the link to see
how the bug was solved.


> "GbTS" == GNU bug Tracking System  writes:
GbTS> This is an automatic notification regarding your bug report
GbTS> #14971: split man page table mushed,
GbTS> which was filed against the coreutils package.

GbTS> Thank you for your report, which has now been closed.
GbTS> You can view the full report at
GbTS> http://debbugs.gnu.org/cgi/bugreport.cgi?bug=14971

GbTS> If you require further information, please followup to 
14...@debbugs.gnu.org.

GbTS> debbugs.gnu.org maintainers
GbTS> (administrator, GNU bugs database)

bug#18328: can't say date -d '8pm -0500' though other combos work

[積丹尼 Dan Jacobson]

AG> I hope to get to this bug soon.

Good.

bug#33025: Add examples of why one would want to "sort" something "randomly"

[積丹尼 Dan Jacobson]

On (info "(coreutils) sort invocation")

‘-R’
‘--random-sort’
‘--sort=random’
 Sort by hashing the input keys and then sorting the hash values.
 Choose the hash function at random, ensuring that it is free of
 collisions so that differing keys have differing hash values.  This
 is like a random permutation of the inputs (*note shuf
 invocation::), except that keys with the same value sort
 together...

OK, but you need to mention some examples of why someone would want to
"sort" something "randomly".

OK, I have one:

You have a list of URLs.

You know there are probably duplicates in there, that you want to get
rid of.

But you also don't want to make it look like you have "machine sorted
them." You want them to still look "hand assembled."

So you use sort --random-sort --unique on them!

bug#31055: Document ls -Ur -fr

[積丹尼 Dan Jacobson]

(info "(coreutils) Sorting the output") says

‘-r’
‘--reverse’
 Reverse whatever the sorting method is—e.g., list files in reverse
 alphabetical order, youngest first, smallest first, or whatever.

OK but mention 'whatever' doesn't mean -U, -f, because they are sorted
in "unsorted" order. In fact perhaps trigger an error.

Actually directory order is still an order, so in fact it is odd that -r
is "broken" here.

bug#31038: mv copies in ls -r order

[積丹尼 Dan Jacobson]

OK. Please on (info "(coreutils) cp invocation") at
‘-v’
‘--verbose’
 Print the name of each file before copying it.
add:

"(Note this reveals internal order of operations, often in ls -U order,
thus occasionally appearing to be in ls, ls -t, ls -r, ls -rt etc. order.)"

(I mean on the same page you do go into such details:

‘-a’
‘--archive’
 Preserve as much as possible of the structure and attributes of the
 original files in the copy (but do not attempt to preserve internal
 directory structure; i.e., ‘ls -U’ may list the entries in a copied
 directory in a different order).  Try to preserve SELinux security
 context and extended attributes (xattr), but ignore any failure to
 do that and print no corresponding diagnostic.  Equivalent to ‘-dR

)

bug#31038: mv copies in ls -r order

[積丹尼 Dan Jacobson]

OK maybe I was just looking at the latter half of

$ sh O
/tmp
created directory '/var/tmp/y'
copied 'x/1' -> '/var/tmp/y/1'
copied 'x/2' -> '/var/tmp/y/2'
copied 'x/3' -> '/var/tmp/y/3'
copied 'x/4' -> '/var/tmp/y/4'
copied 'x/5' -> '/var/tmp/y/5'
copied 'x/6' -> '/var/tmp/y/6'
copied 'x/7' -> '/var/tmp/y/7'
copied 'x/8' -> '/var/tmp/y/8'
copied 'x/9' -> '/var/tmp/y/9'
removed 'x/9'
removed 'x/8'
removed 'x/7'
removed 'x/6'
removed 'x/5'
removed 'x/4'
removed 'x/3'
removed 'x/2'
removed 'x/1'
removed directory 'x'

$ cat O
set -eu
cd /tmp
mkdir x
cd x
seq 9|xargs touch
cd -
mv -v x /var/tmp/y

bug#31038: mv copies in ls -r order

[積丹尼 Dan Jacobson]

$ mv -v dir1/* dir2
reveals that mv works backwards,
copying in ls -r order.
Well OK, but why is that order better than the order of the arguments it
was given?

bug#29604: man cut should mention no way to do ls -l|cut ...

[積丹尼 Dan Jacobson]

> "EB" == Eric Blake  writes:

EB> Let's step back a bit. We recommend that you NEVER parse ls output, as

No problem. Any example of what can't be parsed would be fine. Even e.g.,

EB> Eric Blake, Principal Software Engineer
EB> Red Hat, Inc.   +1-919-301-3266
EB> Virtualization:  qemu.org | libvirt.org

your signature, with differing lengths of whitespace. Whatever.

bug#29604: man cut should mention no way to do ls -l|cut ...

[積丹尼 Dan Jacobson]

On (info "(coreutils) cut invocation")

‘-d INPUT_DELIM_BYTE’
‘--delimiter=INPUT_DELIM_BYTE’
 With ‘-f’, use the first byte of INPUT_DELIM_BYTE as the input
 fields separator (default is TAB).

please mention there is no way to indicate more than one character,
thus no way to process
$ ls -l | cut ...
and that one should use awk, etc. for that.

Also mention it on the cut(1) man page.

bug#29044: sort --debug results improvement

[積丹尼 Dan Jacobson]

< P.S., Yes indeed I had LC_COLLATE=C so maybe --debug should mention
< where in the environment it made it choices from too.

Ah, like you said

  $ LC_ALL=en_CA.UTF-8 sort --debug < /dev/null
  sort: using ‘en_CA.UTF-8’ sorting rules

  $ LC_ALL=C sort --debug < /dev/null
  sort: using simple byte comparison

So the last line should be
  sort: using 'C' sorting rules (simple byte comparison)

or maybe also say "effective LC_COLLATE value is "..

bug#29044: sort --debug results improvement

[積丹尼 Dan Jacobson]

Your answer is absolutely pure gold for a new page linked from

‘--debug’
 Highlight the portion of each line used for sorting.  Also issue
 warnings about questionable usage to stderr.

in the Info manual! Please don't let it go to waste sitting in the bug
tracker. Perhaps call it Debugging examples. You can pretty much just
quote the entire exchange between you and me.

P.S., Yes indeed I had LC_COLLATE=C so maybe --debug should mention
where in the environment it made it choices from too.

bug#28735: say "by the current user"

[積丹尼 Dan Jacobson]

On (info "(coreutils) Access permission tests")
don't say
‘-r FILE’
 True if FILE exists and read permission is granted.

‘-w FILE’
 True if FILE exists and write permission is granted.

‘-x FILE’
 True if FILE exists and execute permission is granted (or search
 permission, if it is a directory).

Say ... "is X-able by the current user."
Same on the ls man page.

Else you are not being clear if this is just testing if a certain bit
has been set, vs. something to do with the current user vs. the file owner!

bug#27786: tr: accept hex ranges as well as octal

[積丹尼 Dan Jacobson]

In (info "(coreutils) Character sets")

 ‘\OOO’
  The 8-bit character with the value given by OOO, which is 1 to
  3 octal digits.  Note that ‘\400’ is interpreted as the
  two-byte sequence, ‘\040’ ‘0’.

OK but do add hex too. Hmm, I probably reported this before but can't
tell because I'm offline while I'm writing this, and it is too important
anyway for me to check vs. mentioning again.

bug#27785: tr: mention that leading zeros are a no-no in e.g., \0377

[積丹尼 Dan Jacobson]

In (info "(coreutils) Character sets")

 ‘\OOO’
  The 8-bit character with the value given by OOO, which is 1 to
  3 octal digits.  Note that ‘\400’ is interpreted as the
  two-byte sequence, ‘\040’ ‘0’.

It is very important to also add here:

  and \0377 is interpreted as \037 and 7.

(You add the quotes. I can't with my keyboard.)

bug#27482: date: minus and ago in same sentence

[積丹尼 Dan Jacobson]

I got an idea.
Please parse "one", "two", "three" (lower and upper case folded too.)
1. Because it is proper English. "1 day ago" is cellphone English.
2. Because there can't be that many of them... (seventy-three,
seventy-four... well they are made up of other components...)
3. Because it will avoid the user getting these surprising results,
$ date -d 'two weeks ago'
date: invalid date ‘two weeks ago’
$ date -d 'now - two weeks ago'
date: invalid date ‘now - two weeks ago’
$ date -d 'now - 2 weeks ago'
Sun Jul  9 18:39:23 CST 2017

Or just catch the error of "-" and "ago" in the same sentence.

Or perhaps warn on (info "(coreutils) Relative items in date strings")
about minus and ago...

bug#26971: mv -v output ordering looks impossible

[積丹尼 Dan Jacobson]

I don't think you guys should invent new eye-catching strings.
Aren't there already standard ways to show what is going on under the
hood? Let's see, sh -x uses + and ++... strace uses... maybe just use '# '
comments.

bug#26971: mv -v output ordering looks impossible

[積丹尼 Dan Jacobson]

I'm fine with whatever you guys come up with, just don't
R> - Leave the output like it was initially.
else I'll just be back here five years later after forgetting the whole episode.

bug#26971: mv -v output ordering looks impossible

[積丹尼 Dan Jacobson]

PB> So this is working across file systems
Yes.
>> '/home/jidanni/jidanni.org/location/grow/programs' -> '/tmp/programs'
This says to me "I, the mv command, have just moved  A to B.

PB> create dir /tmp/programs
If that (create B) is what it is doing in that step, then it should not
mention the unrelated A.

>> '/home/jidanni/jidanni.org/location/grow/programs/grow.tgz' -> 
>> '/tmp/programs/grow.tgz'
This says to me "I am moving A/C to B/C". But there is no more A... at
least that is what the user thinks... so how could it still move it.


PB> I suppose we could distinguish the creation operations, though
PB> note if the first attempted rename() worked for the source dir,
PB> then there would have been just a single operation output like:

PB>   '/home/jidanni/jidanni.org/location/grow/programs' -> '/tmp/programs'

Well all I know is if the user always can do something on one line, then he
would expect a consistent number of -v output lines.

bug#26971: mu -v output ordering looks impossible

[積丹尼 Dan Jacobson]

I do
mv -v /home/jidanni/jidanni.org/location/grow/programs /tmp
and see
'/home/jidanni/jidanni.org/location/grow/programs' -> '/tmp/programs'
'/home/jidanni/jidanni.org/location/grow/programs/grow.tgz' -> 
'/tmp/programs/grow.tgz'
removed '/home/jidanni/jidanni.org/location/grow/programs/grow.tgz'
removed directory '/home/jidanni/jidanni.org/location/grow/programs'
but isn't that a rather impossible sequence? At least the first two?
Well I don't know the best way to write it.

bug#26902: good invalid dates vs. bad invalid dates

[積丹尼 Dan Jacobson]

Can you please don't say "invalid date" for everything invalid.

$ date -d @1494439601675485
date: invalid date ‘@1494439601675485’
$ date -d 1494439601675485sec
date: invalid date ‘1494439601675485sec’
$ date -d 1494439601675485seconds
date: invalid date ‘1494439601675485seconds’
$ date -d @1494439601675485
date: invalid date ‘@1494439601675485’
$ date -d @1494439601675
date: invalid date ‘@1494439601675’
$ date -d @1
四  1月  1 08:00:01 CST 1970
$ date -d @149443960167
date: invalid date ‘@149443960167’
$ date -d @14944396016
date: invalid date ‘@14944396016’
$ date -d @1494439601
四  5月 11 02:06:41 CST 2017
$ date -d @zz
date: invalid date ‘@zz’

On (info "(coreutils) Seconds since the Epoch") it says

  Traditional Unix systems count seconds with 32-bit two’s-complement
   integers and can represent times from 1901-12-13 20:45:52 through
   2038-01-19 03:14:07 UTC.  More modern systems use 64-bit counts of
   seconds with nanosecond subcounts, and can represent all the times in

Please catch the cases where the user merely has entered too many digits
for the computer in question, and say
date: too far into the future.
or
date: too far into the future for current computer.
or
date: invalid date: too far into the future for current computer.

Please don't just give the same message that you give if the user hasn't
even entered one digit.

bug#26631: tr: extra operand ‘\\\\’

[積丹尼 Dan Jacobson]

$ tr -s x -d '\\'
tr: extra operand ‘’
Try 'tr --help' for more information.

I only gave '\\'.

bug#25870: give an example to show what lines join -a and -v are talking about

[積丹尼 Dan Jacobson]

I can't really tell, but it seems at the beginning there still lacks the
most basic "Hi kids, here is the 'join' command. It takes one file with
contents ... and and another file with contents ... and outputs ..."
before you get into any more niceties.

bug#25870: give an example to show what lines join -a and -v are talking about

[積丹尼 Dan Jacobson]

In (info "(coreutils) join invocation")

‘-a FILE-NUMBER’
 Print a line for each unpairable line in file FILE-NUMBER (either
 ‘1’ or ‘2’), in addition to the normal output.

OK but say if we can use both: -a 1 -a 2? (Answer: yes)
And if so how to tell which lines are from which file (Answer: by their content 
only).

‘-v FILE-NUMBER’
 Print a line for each unpairable line in file FILE-NUMBER (either
 ‘1’ or ‘2’), instead of the normal output.

OK but say if we can use both: -v 1 -v 2?
And if so how to tell which lines are from which file.

And in fact give an example to show what lines -a and -v are talking about!

bug#25871: give plain join example first

[積丹尼 Dan Jacobson]

In (info "(coreutils) join invocation")

   If the input has no unpairable lines, a GNU extension is available;

(shouldn't that ";" be ":"?)


the sort order can be any order that considers two fields to be equal if
and only if the sort comparison described above considers them to be
equal.  For example:

 $ cat file1...

OK, but perhaps in the previous paragraph give a mundane regular
example of join usage, instead of the first example be this special
exception.

bug#25833: fold -s ... but for more than just spaces

[積丹尼 Dan Jacobson]

fold -s will break at spaces.
But why not let the user also pick an arbitrary character, other than
spaces, too?

bug#25745: make dd quiet except for warnings and errors

[積丹尼 Dan Jacobson]

That was Info dd. man dd says
   status=LEVEL
  The  LEVEL  of information to print to stderr; 'none' suppresses
  everything but error messages,  'noxfer'  suppresses  the  final
  transfer  statistics, 'progress' shows periodic transfer statis-
  tics

And when you think about it, the

0+0 records in
0+0 records out

sure look like part of something called final transfer statistics...