naren@clan.UUCP (Narendra Mehta) (08/18/86)
grep '[' file
gives a 'RE botch' error on our VAX-750 running 4.2BSD. If the fix
has already been posted, could someone please mail me the diffs? Thanks.
------------------------------------------------------------------
Narendra Mehta,
Systems & Computer Engineering Department, Carleton University, Ottawa
UUCP: {allegra,decvax,ihnp4,linus,utzoo}!watmath!clan!narenbzs@bu-cs.BU.EDU (Barry Shein) (08/21/86)
> grep '[' file > > gives a 'RE botch' error on our VAX-750 running 4.2BSD. If the fix >has already been posted, could someone please mail me the diffs? Thanks. >------------------------------------------------------------------ >Narendra Mehta, That's not a bug, '[' is special to grep, try '\[' to escape it. -Barry Shein, Boston University
stephen@datacube.UUCP (08/22/86)
> > grep '[' file > > gives a 'RE botch' error on our VAX-750 running 4.2BSD. If the fix > has already been posted, could someone please mail me the diffs? Thanks. > ------------------------------------------------------------------ > Narendra Mehta, > Systems & Computer Engineering Department, Carleton University, Ottawa > UUCP: {allegra,decvax,ihnp4,linus,utzoo}!watmath!clan!naren Try grep '\[' file. Stephen Watkins UUCP: ihnp4!datacube!stephen Datacube Inc.; 4 Dearborn Rd.; Peabody, Ma. 01960; 617-535-6644
john@uw-nsr.UUCP (John Sambrook) (08/22/86)
In article <191@clan.UUCP> naren@clan.UUCP (Narendra Mehta) writes: > > grep '[' file > > gives a 'RE botch' error on our VAX-750 running 4.2BSD. If the fix >has already been posted, could someone please mail me the diffs? Thanks. The message is correct. On our system it is a little clearer: grep: RE error 49: [ ] imbalance. The "[" character is used to introduce a range specification, as in "[a-z]" to match a single character in the range of "a" to "z". To search for a literal '[' character you should escape it with a backslash, as in: grep '\[' file Hope this helps. -- John Sambrook Work: (206) 545-2018 University of Washington WD-12 Home: (206) 487-0180 Seattle, Washington 98195 UUCP: uw-beaver!uw-nsr!john
jsloan@wright.EDU (John Sloan) (08/22/86)
> > grep '[' file > > > > gives a 'RE botch' error on our VAX-750 running 4.2BSD. > >Narendra Mehta, > That's not a bug, '[' is special to grep, try '\[' to escape it. > -Barry Shein, Boston University There is still a bug though: while [ is a special pattern operator to grep, a single [ as in the example causes my 4.2BSD grep to go into a loop issuing the "RE botch" error over and over. So while the error message is legitimate, the ungraceful nonrecovery is not. -- ------------------------------------------------------------------------ John Sloan CSNET: jsloan@WRIGHT.EDU Wright State University USENET: ...!cbosgd!wright!jsloan Computer Science and Engineering ARPANET: jsloan%wright@CSNET-RELAY Dayton, OH 45435 DECNET LAN: wright::jsloan +1 513 873 2491, 2987 SMTP LAN: jsloan@wright UNIX is a registered trademark of AT&T. Opinions expressed are my own. "I was a roadie for the Algorithmics on their 1986 Internet Tour"
dce@mips.UUCP (David Elliott) (08/23/86)
In article <191@clan.UUCP> naren@clan.UUCP (Narendra Mehta) writes: > > grep '[' file > > gives a 'RE botch' error on our VAX-750 running 4.2BSD. If the fix >has already been posted, could someone please mail me the diffs? Thanks. The fix is to put 'exit(2)' after the line fprintf(stderr, "RE botch\n"); in compile(), and the line fprintf(stderr, "RE error\n"); in advance(). This only applies to /usr/ucb/grep in 4.2 and 4.3. Without these, you get an 'RE error' message followed by a whole slew of 'RE botch' messages. There isn't any reason for grep to keep going if the RE compiles wrong, anyway. Admittedly, when I first saw this bug reported, I fired off a note to Narendra saying it wasn't a bug. Like a fool (sorry Barry), I didn't try it out first (I did this last year as well, an event Dave Sherman may remember). David
ado@elsie.UUCP (Arthur David Olson) (08/26/86)
Index: ucb/grep 4.1 Fix Description: > There is still a bug though: while [ is a special pattern operator to grep, > a single [ as in the example causes my 4.2BSD grep to go into a loop issuing > the "RE botch" error over and over. So while the error message is legitimate, > the ungraceful nonrecovery is not. Repeat-By: grep \[ /etc/passwd Fix: Is to "ucbgrep.c" As usual, the trade secret status of the code involved precludes a clearer posting. The fix is the addition conditioned on "OLDVERSION" below; it goes at the end of the function "compile". cerror: fprintf(stderr, "RE error\n"); #ifndef OLDVERSION exit(2); #endif } -- Bug/s is a Volkswagen/Warner Brothers trademark. Wright is a Safeway trademark. -- UUCP: ..decvax!seismo!elsie!ado ARPA: elsie!ado@seismo.ARPA DEC, VAX, Elsie & Ado are Digital, Borden & Ampex trademarks.
lepreau@utah-cs.UUCP (Jay Lepreau) (08/29/86)
On the contrary, this is fixed in the 4.3 release. You must be looking at 4.3 Beta.
dce@mips.UUCP (David Elliott) (09/01/86)
In article <3907@utah-cs.UUCP> lepreau@utah-cs.UUCP (Jay Lepreau) writes: >On the contrary, this is fixed in the 4.3 release. You >must be looking at 4.3 Beta. Sorry, but we have standard 4.3 and it's broken. You must understand that the problem is in /usr/ucb/grep and not in /bin/grep in 4.3. In other words, it's only half fixed. BTW, why are there two greps in 4.x? David