On Thu, Jun 14, 2007 at 11:27:02AM -0400, Josef Sipek wrote: > On Thu, Jun 14, 2007 at 02:50:55PM +0200, Pierre Habouzit wrote: > > This patch series makes guilt be able to work with a standard posix shell. > > > > This has been tested with bash, zsh, dash and posh acting as /bin/sh. > > > > This uses awk and sed a bit more than before, but POSIX awk and sed should be > > enough. Though GNU find and GNU /bin/echo from the coreutils needs to be > > there. > > > > Here is the shortlog: > > > > Pierre Habouzit (7): > > Regression test suite needs bash, that's OK. > > guilt(1): Obvious bashisms fixed. > > guilt(1): simplifications... > > guilt(1): reimplement push_patch, using a subshell to avoid locals. > > Easy commands, without bashisms. > > guilt-status(1): Remove bashisms. > > Remove last bashisms from remaining commands. > > I went though the patches, and it's mostly good (see individual patch > replies for details). I noticed an indentation issue or two, but I can fix > that up easily, I can take care of the quoting of `basename $0`. Thanks. > The regression suite is little lacking, so I am going to put your updated > patches onto a special branch, and test it for a while - fixing up any > potential bugs. (I don't want to be even remotely responsible for people's > patches turning into empty files :) ) That seems like a very good plan. I just sent the fixes. Hopefully this should fix major issues right now. Interestingly enough, benchmarking for runs with different shells as /bin/sh show: dash: 1.63s user 3.93s system 102% cpu 5.436 total posh: 2.02s user 4.29s system 100% cpu 6.260 total bash: 2.82s user 5.78s system 104% cpu 8.223 total zsh: 3.70s user 5.86s system 99% cpu 9.626 total (this is obviously completely not unreliable, but gives some rough ideas though) > Thanks a lot! Well, worry for the obvious mistakes. -- ·O· Pierre Habouzit ··O madcoder@xxxxxxxxxx OOO http://www.madism.org
Attachment:
pgpg8RuLQkmdd.pgp
Description: PGP signature
