GNU bug report logs - #33162
Pattern partially greps windows line-endings since 3.1

Previous Next

To reply to this bug, email your comments to 33162 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: "Holger Klene" <h.klene <at> gmx.de>
To: bug-grep <at> gnu.org
Subject: Pattern partially greps windows line-endings since 3.1
Date: Fri, 26 Oct 2018 14:41:59 +0200

Hello!

I think, I may have stumbled upon a bug in gnu-grep version 3.1 which was not present in 3.0 and 2.6.3

Steps to reproduce:
A) generate test-file: printf "a --\r\nb --\nc --\rd --\n\r" > test.txt

B) check test-file: md5sum.exe test.txt
e3ef5d742eee3a7f326bc7b4b391cc42 *test.txt

C) run old grep 3.0 or 2.6.3:
$ grep -ce '--[^ -]' test.txt
1
$ grep -ne '--[^ -]' test.txt
3:c --d --

D) run new grep 3.1:
$ grep -ce '--[^ -]' test.txt
2
$ grep -ne '--[^ -]' test.txt

d --

It looks as if the 3.1 has an issue with alternative line-ending styles sometimes matched as part of the pattern:
a --\r\n    (windows) is messed around
b --\n      (unix) seems to work OK
c --\r      just for completenes
d --\n\r    just for completenes

You can modify the printf statement to generate each test-case into it's own file. b and d are equally not found, c is always found. The behavior changed for a: [^ -] now matches the \r in 3.1 which was ignored in older versions.

Thanks
Holger

PS: This version of grep was distributed along with git-bash for Windows 64bit
https://github.com/git-for-windows/git/releases/download/v2.19.1.windows.1/Git-2.19.1-64-bit.exe
grep (GNU grep) 3.1
GNU bash, version 4.4.19(2)-release (x86_64-pc-msys)
git version 2.19.1.windows.1

Message #12 received at 33162 <at> debbugs.gnu.org (full text, mbox):

From: Paul Eggert <eggert <at> cs.ucla.edu>
To: Konstantin Kolinko <knst.kolinko <at> gmail.com>, Holger Klene <h.klene <at> gmx.de>
Cc: 33162 <at> debbugs.gnu.org, 32838 <at> debbugs.gnu.org
Subject: Re: bug#32838: Regression in grep 3.1: Grep '$' matcher no longer
 works on CRLF files on Windows
Date: Mon, 21 Sep 2020 15:41:23 -0700

On 9/25/18 1:15 AM, Konstantin Kolinko wrote:

> https://github.com/git-for-windows/git/issues/1841
Looking at that downstream bug report, it appears that this is a wishlist 
request and that there have been no takers for implementing an 
MS-Windows-specific --strip-trailing-cr option. I have marked Bug#32838 and 
Bug#33162 as wishlist in the grep bug database.

Previous Next

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