svn commit: r258848 - user/hrs/releng/release

[Hiroki Sato]

Glen Barber <gjb at FreeBSD.org> wrote
  in <20131202160333.GI1839 at glenbarber.us>:

gj> >  - In the current release.sh, variables unintentionally set in the
gj> >    builder's environment still pollutes commands which run in
gj> >    release.sh when the variables are not set in the script but
gj> >    accepted by make or other utilities.  Catching all of the variables
gj> >    is quite difficult.
gj> >
gj>
gj> Do you have an example?  If you are referring to a previous problem with
gj> 10.x builds, the actual problem was a race in the build toolchain, and
gj> not variable pollution.  I was able to confirm this was the case there.

 One of the surprising examples I experienced was that svn co
 sometimes fails with LC_ALL=C.  Subversion stores repo data in UTF-8,
 so iconv conversion happens when LC_ALL is not UTF-8 compatible (and
 it broke "svn co head/" a while ago).  Besides, TARGET, TARGET_ARCH,
 MAKEOBJDIR, MAKEOBJDIRPREFIX, and OBJDIR are also dangerous and
 unprotected variables for make(1).  An empty TARGET caused an
 infinite loop in libc build, so I fixed it in Makefile yesterday.
 All of them are examples of "unintentional" variables, so being
 careful is enough to avoid them, though.

gj> >  Anyway, I do not have a strong opinion if using FOO=A or : ${FOO:=A}.
gj> >  I agree that the latter assumes all of the envvars must be under
gj> >  control and it needs one more step than just invoking the script, and
gj> >  a leaked variable can break the build.  If you do not like it, I will
gj> >  change them with FOO=A.
gj> >
gj>
gj> Yes, please.

 Will do.  I will not merge the changes before 10.0R is out in any
 case.

-- Hiroki
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 196 bytes
Desc: not available
URL: <http://lists.freebsd.org/pipermail/svn-src-user/attachments/20131203/4ca22732/attachment.sig>