GNU bug report logs - #27136
doc: cat: explain what is "standard input"

Previous Next

To reply to this bug, email your comments to 27136 AT debbugs.gnu.org.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: kalle <kalle <at> projektwerkstatt.de>
To: bug-coreutils <at> gnu.org
Subject: info coreutils
Date: Mon, 29 May 2017 23:06:23 +0200

[Message part 1 (text/plain, inline)]

[bug-coreutils@gnu.org (text/plain, attachment)]

Message #8 received at 27136 <at> debbugs.gnu.org (full text, mbox):

From: Pádraig Brady <P <at> draigBrady.com>
To: kalle <kalle <at> projektwerkstatt.de>, 27136 <at> debbugs.gnu.org
Subject: Re: bug#27136: info coreutils
Date: Mon, 29 May 2017 15:08:27 -0700

On 29/05/17 14:06, kalle wrote:
> hello,
> I think it is not obvious for everyone how to deal with Standard Input in the `cat'-command. I took some time to figure out, that Ctrl-d  aborts it. and just now I realized, that EOF the first time acts to terminate a so called "record" (I got the term from "info coreutils 'tac'"), and applied twice terminates the Standard Input. If <newline> was typed before it is enough to type it once.
> So I think it would be helpful for some people if in the documentations man/info cat there would be at least a reference, if not a short explanation how to deal with Standard Input.

Note this is nothing specific to cat or coreutils for that matter.
This applies to all programs and is specific to the terminal driver.
Now where info like this should be documented is an interesting question.
Perhaps coreutils should take some responsibility for documenting such things?

Anyway some info for what is happening here...
Note there is no "record" distinction here.
The Ctrl-d combo just means "end input". I.E. send pending input to the program.
When there is text on the line already this is sent, and usually (as cat does too)
the program will read() again looking for more input. If you send pending input
on an empty line, then the program will get zero bytes and usually quit.
You can see this with:

  strace -e read cat

Note the Ctrl-d combo can be changed in the terminal driver
using the slightly misnamed eof stty setting:

  $ stty -a | grep eof
  intr = ^C; quit = ^\; erase = ^?; kill = ^U; eof = ^D; eol = <undef>;

cheers,
Pádraig

Message #11 received at 27136 <at> debbugs.gnu.org (full text, mbox):

From: kalle <kalle <at> projektwerkstatt.de>
To: Pádraig Brady <P <at> draigBrady.com>, 27136 <at> debbugs.gnu.org
Subject: Re: bug#27136: info coreutils
Date: Sun, 13 Aug 2017 12:22:46 +0200

hmm. no one is responding anymore. No change in cat-Documentation then?

Am 25.07.2017 um 00:31 schrieb kalle:
> 
> 
> Am 30.05.2017 um 00:08 schrieb Pádraig Brady:
>> On 29/05/17 14:06, kalle wrote:
>>> hello,
>>> I think it is not obvious for everyone how to deal with Standard Input in the `cat'-command. I took some time to figure out, that Ctrl-d  aborts it. and just now I realized, that EOF the first time acts to terminate a so called "record" (I got the term from "info coreutils 'tac'"), and applied twice terminates the Standard Input. If <newline> was typed before it is enough to type it once.
>>> So I think it would be helpful for some people if in the documentations man/info cat there would be at least a reference, if not a short explanation how to deal with Standard Input.
>>
>> Note this is nothing specific to cat or coreutils for that matter.
>> This applies to all programs and is specific to the terminal driver.
> 
>> Now where info like this should be documented is an interesting question.
> 
> maybe in a general unix-introduction. Then for all commands, where "end
> input" has to be given in instead of Enter/linefeed, it should  be
> referred to some explaining text. There could be a manual about the role
> of terminal drivers, possibly man 7.
> 
>> Perhaps coreutils should take some responsibility for documenting such things?
>>
>> Anyway some info for what is happening here...
>> Note there is no "record" distinction here.
>> The Ctrl-d combo just means "end input". I.E. send pending input to the program.
>> When there is text on the line already this is sent, and usually (as cat does too)
>> the program will read() again looking for more input. If you send pending input
>> on an empty line, then the program will get zero bytes and usually quit.
>> You can see this with:
>>
>>   strace -e read cat
>>
>> Note the Ctrl-d combo can be changed in the terminal driver
>> using the slightly misnamed eof stty setting:
>>
>>   $ stty -a | grep eof
>>   intr = ^C; quit = ^\; erase = ^?; kill = ^U; eof = ^D; eol = <undef>;
>>
>> cheers,
>> Pádraig
>>

Message #14 received at 27136 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: kalle <kalle <at> projektwerkstatt.de>, 27136 <at> debbugs.gnu.org
Subject: Re: bug#27136: info coreutils
Date: Sun, 13 Aug 2017 11:22:49 -0700

kalle wrote:
> no one is responding anymore. No change in cat-Documentation then?

Nothing so far, yes. No specific wording changes have been proposed, as far as I 
can see from <https://bugs.gnu.org/27136>. You can propose a wording change by 
sending us the output of "diff -u" or "git send-email" or something like that. 
It's OK if you don't have time to do that, as this is evidently low-priority.

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.