Path: csiph.com!goblin2!goblin1!goblin.stu.neva.ru!usenet.stanford.edu!not-for-mail From: Greg Wooledge Newsgroups: gnu.bash.bug Subject: Re: Word boundary anchors \< and \> not parsed correctly on the right side of =~ Date: Tue, 10 Jul 2018 09:09:06 -0400 Lines: 38 Approved: bug-bash@gnu.org Message-ID: References: <5b440fe8.1c69fb81.948f6.4d1e@mx.google.com> <20180710122718.wrrzjuqpmpns4o4o@eeg.ccf.org> <20180710130234.r5mwjc62adids7m5@monk> NNTP-Posting-Host: lists.gnu.org Mime-Version: 1.0 Content-Type: text/plain; charset=iso-8859-1 Content-Transfer-Encoding: 8bit X-Trace: usenet.stanford.edu 1531228156 29795 208.118.235.17 (10 Jul 2018 13:09:16 GMT) X-Complaints-To: action@cs.stanford.edu Cc: bug-bash@gnu.org, bash@packages.debian.org To: Paulo Marcel Coelho =?iso-8859-1?Q?Arag=E3o?= Envelope-to: bug-bash@gnu.org Mail-Followup-To: Paulo Marcel Coelho =?iso-8859-1?Q?Arag=E3o?= , bug-bash@gnu.org, bash@packages.debian.org Content-Disposition: inline In-Reply-To: <20180710130234.r5mwjc62adids7m5@monk> User-Agent: NeoMutt/20170113 (1.7.2) X-detected-operating-system: by eggs.gnu.org: GNU/Linux 2.2.x-3.x [generic] [fuzzy] X-Received-From: 139.137.100.1 X-BeenThere: bug-bash@gnu.org X-Mailman-Version: 2.1.21 Precedence: list List-Id: Bug reports for the GNU Bourne Again SHell List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Xref: csiph.com gnu.bash.bug:14310 On Tue, Jul 10, 2018 at 10:02:34AM -0300, Paulo Marcel Coelho Aragão wrote: > It baffles me that literal \< \> are not evaluated correctly > > paulo:~$ [[ 'foo bar' =~ \ ]] && echo yes || echo no > no In this context, the backslashes serve only to "quote" the less-than and greater-than signs. It's just like writing [[ 'foo bar' =~ "<"foo">" ]] And when you quote a character on the right-hand side of = or =~ inside a [[ command, that character LOSES whatever specialness it might have had in that context. In order to put a backslash-that-acts-like-a-backslash-to-the-regex-engine inside the RE on the right-hand side of =~ you need to put the RE into a variable. (Normally this is not a problem because normally you write POSIX ERE-compliant regexes because that's all that you can count on, and POSIX EREs don't use backslashes to turn features ON. Only to turn them OFF.) > Empirically all this demonstrates: > > 1. bash recognizes \< \> \b as word anchors Only on some systems. Not on others. So don't do that. > 2. bash doesn't evaluate them correctly when used as literals > 3. bash evaluates them correctly when used with parameter expansion Which is one of the reasons it is STRONGLY encouraged to put your RE into a variable and then use =~ $var. (Another reason is backward compatibility with bash 3.x versions, should you happen to care about that.)