2008/12/19 karl williamson <pub...@khwilliamson.com>: > Craig A. Berry wrote: >> >> On Fri, Dec 19, 2008 at 11:24 AM, Steve Hay <steve...@planit.com> wrote: >>> >>> Rafael Garcia-Suarez wrote: >>>> >>>> 2008/12/19 karl williamson <pub...@khwilliamson.com>: >>>>> >>>>> Two things: when I do as above using rsync, I merely get a list of >>>>> the valid perl sources. If I change the :: to a slash, I actually >>>>> get the files. However, .patch contains >>>>> "blead 20081219.17:05:08 b1fccde5a5d6d9eece3886ce5cb9dc7b9027bee2" >>>>> which doesn't compile well. >>>> >>>> What is exactly the problem with that ? I don't think anything relies >>>> on .patch being an integer, but I might be wrong. >>> >>> Nothing wrong with it here. I did a little smoke test using the new >>> rsync source earlier, and it built fine (see attached). >> >> I think that's because Test::Smoke did it's own editing of >> patchlevel.h. It didn't blow up, but it also didn't really work. >> You'll notice all of the hex digits have been stripped from your SHA1. >> Since the timestamp is embedded in there it's reasonably safe but not >> entirely correct. >> >> > It's the spaces that cause the problem. If I change them to dashes it > works. > > Here's a command line after the spaces have been changed. > `sh cflags "optimize='-O0 -g3'" op.o` op.c > CCCMD = cc -DPERL_CORE -c > -DPERL_PATCHNUM=blead-20081219.17:05:08-b1fccde5a5d6d9eece3886ce5cb9dc7b9027bee2 > -DDEBUGGING -fno-strict-aliasing -pipe -fstack-protector > -I/usr/local/include -D_LARGEFILE_SOURCE -D_FILE_OFFSET_BITS=64 -std=c89 -O0 > -g3 -Wall -ansi -W -Wextra -Wdeclaration-after-statement -Wendif-labels > -Wc++-compat >
I see. I'll change them to dashes then. But anyway Test::Smoke should be amended to cope with non-digits.