GNU bug report logs - #75197
[Feature Request] Support 'grep -A' without number to show all lines after match

Previous Next

To reply to this bug, email your comments to 75197 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: KMRH <kmrh47 <at> proton.me>
To: "bug-grep <at> gnu.org" <bug-grep <at> gnu.org>
Subject: [Feature Request] Support 'grep -A' without number to show all lines
 after match
Date: Mon, 30 Dec 2024 11:37:27 +0000

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

Package: grep
Severity: wishlist

Dear Maintainers,

I would like to request a feature enhancement for the grep tool.

Feature Request
Allow the command 'grep -A' (without specifying a number) to default to showing all lines after a match, instead of erroring out. This would be conceptually equivalent to something like using a very large number ('grep -A 999999 <pattern>'), but without having to manually pick a placeholder length.

Current Behavior
$ grep -A <pattern>
grep: <pattern>: Invalid context length argument

Proposed Behavior
$ grep -A <pattern> # show all lines after match

Rationale

- Intuitive defaults: If no numeric limit is specified, show everything after the match.

- Cleaner syntax: Avoids the arbitrary use of large numbers.
- Unix philosophy: Simple, predictable defaults that reduce friction.
- Alternative approach: If there’s concern about backward compatibility, consider a more explicit flag like '-A all'

I acknowledge that some scripts may rely on the current error behavior for an empty '-A' argument. However, I believe that in most cases it would not cause significant breakage. If necessary, we could introduce a distinct option to avoid disrupting existing usage.

Thank you for considering this enhancement.

Best regards,
Karsten Holmøe

[Message part 2 (text/html, inline)]

Message #8 received at 75197 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: KMRH <kmrh47 <at> proton.me>
Cc: 75197 <at> debbugs.gnu.org
Subject: Re: bug#75197: [Feature Request] Support 'grep -A' without number to
 show all lines after match
Date: Mon, 30 Dec 2024 10:52:53 -0800

On 2024-12-30 03:37, KMRH via Bug reports for GNU grep wrote:
> Allow the command 'grep -A' (without specifying a number) to default to showing all lines after a match

We couldn't do that as-is, as 'grep -A 1' is equivalent to 'grep -A1' 
and people rely on that.

'grep -A 99999999999999999999999999999999999999999' does what you want 
already, even if 99999999999999999999999999999999999999999 is too big to 
fit into a machine word. Is that good enough? It would save your having 
to wait for the next grep release....

Alternatively, I suppose we could add syntax like 'grep -A Inf' to do 
what's requested, where "Inf" is short for infinity.

Message #11 received at 75197 <at> debbugs.gnu.org (full text, mbox):

From: arnold <at> skeeve.com
To: kmrh47 <at> proton.me, eggert <at> cs.ucla.edu
Cc: 75197 <at> debbugs.gnu.org
Subject: Re: bug#75197: [Feature Request] Support 'grep -A' without number to
 show all lines after match
Date: Mon, 30 Dec 2024 12:44:24 -0700

Paul Eggert <eggert <at> cs.ucla.edu> wrote:

> Alternatively, I suppose we could add syntax like 'grep -A Inf' to do 
> what's requested, where "Inf" is short for infinity.

And just wait, as soon as you do that, someone will want

	grep -A NaN

:-) :-) :-)

Arnold

Message #14 received at 75197 <at> debbugs.gnu.org (full text, mbox):

From: jackson <at> fastmail.fm
To: "Paul Eggert" <eggert <at> cs.ucla.edu>, KMRH <kmrh47 <at> proton.me>
Cc: 75197 <at> debbugs.gnu.org
Subject: Re: bug#75197: [Feature Request] Support 'grep -A' without number to
 show all lines after match
Date: Mon, 30 Dec 2024 14:14:35 -0600

Or, for us old farts who have been using 'sed' for a half century, if I wanted to print the lines in a file starting with the line matching a certain pattern, through the remaining lines of the file, I'd use:

  sed -n '/pattern/,$/p' < file

Then I wouldn't have to worry whether or not the 'grep' command I was using would lose it's lunch trying to parse '9999999999999999999999', though, granted, every executable 'grep' binary I happen to have on my disk, a couple dozen of them going back over 20 years, all handle the forty-nine 9's I just tested, HOWEVER the busybox 0.51 grep source code circa 2001 I have at hand will fail if strtoul of  the numeric argument to "-A" overflows, returning ULONG_MAX with errno of ERANGE, and likely some older variants of grep would also fail, _and_

I wouldn't have to worry that my 'magic' number wasn't big enough for every ordinary, already written to disk, file that I might be reading, _and_

I could run this off a pipe that never ended with a normal closure of the input stream that would have sent an EOF to input stream of 'sed', but only ended with a signal or hardware stoppage, without worrying that I had not put enough nine's (9) in my number, _and_

I wouldn't have readers of that code wondering what that strange number was ... though, granted, I would have them doing 'man sed' ... about time in my book.

But, yes, as Paul Eggert notes, command line parsing of 'grep -A 123'  would be ambiguous.  Is "123"  a filename, or a parameter to the "-A" argument?

-- 
Paul Jackson
jackson <at> fastmail.fm

Previous Next

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