Re: port-i386/51358: /usr/bin/cvs fakes file permissions

[Richard PALO <richard%netbsd.org@localhost>]

The following reply was made to PR port-i386/51358; it has been noted by GNATS.

From: Richard PALO <richard%netbsd.org@localhost>
To: gnats-bugs%NetBSD.org@localhost
Cc: 
Subject: Re: port-i386/51358: /usr/bin/cvs fakes file permissions
Date: Mon, 01 Aug 2016 19:29:01 +0200

 Le 01/08/16 19:05, Joerg Sonnenberger a écrit :
 >  >  On Thu, Jul 28, 2016 at 05:10:01AM +0000, Richard Palo wrote:
 >  >  >  Okay, I tried a workaround to the wrapper invocation of pkgsrc's missing as follows:
 >  >  >  > nb32$ cvs diff
 >  >  >  > Index: create.mk
 >  >  >  > ===================================================================
 >  >  >  > RCS file: /cvsroot/pkgsrc/mk/tools/create.mk,v
 >  >  >  > retrieving revision 1.5
 >  >  >  > diff -u -r1.5 create.mk
 >  >  >  > --- create.mk   27 Feb 2013 12:09:39 -0000      1.5
 >  >  >  > +++ create.mk   28 Jul 2016 05:01:01 -0000
 >  >  >  > @@ -108,7 +108,7 @@
 >  >  >  >  
 >  >  >  >  .for _t_ in ${TOOLS_GNU_MISSING}
 >  >  >  >  TOOLS_CREATE+=         ${_t_}
 >  >  >  > -TOOLS_PATH.${_t_}?=    ${PKGSRCDIR}/mk/gnu-config/missing
 >  >  >  > +TOOLS_PATH.${_t_}?=    ${TOOLS_SHELL} ${PKGSRCDIR}/mk/gnu-config/missing
 >  >  >  >  TOOLS_SCRIPT.${_t_}?=  ${TOOLS_PATH.${_t_}} ${_t_:T:C/-[0-9].*$//}
 >  >  >  >  .endfor
 >  >  >  >  
 >  >  >  This seems to work in the case of x11/xproto and as well in another case encountered
 >  >  >  textproc/json-c.
 >  >  
 >  >  Fine with me. Probably also needed in texinfo.mk.
 >  
 >  NAK. TOOLS_PATH as the name implies is really meant to be path.
 >  
 >  Joerg
 >  
 > 
 
 In my original test patch, I put ${TOOLS_SHELL} before ${TOOLS_PATH} in TOOLS_SCRIPT.
 
 I opted for the above since given the '?=' (it could possibly be overridden), in which case
 it's not necessarily the pkgsrc 'missing' being discussed.
 
 I agree that this isn't correct given there are other tools that I use TOOLS_PATH for in scripts.
 
 Perhaps either holding off, or putting some execution logic in the wrapper itself if test -x fails
 for POSIX shell scripts (or others) is going to need to be done.
 
 -- 
 Richard PALO