On 25/03/11 05:07, Dan McGee wrote:
On Thu, Mar 24, 2011 at 7:30 AM, Jan Steffens<jan.steffens@gmail.com> wrote:
My main motivation was to remove the "sync", which can stall for minutes on a busy machine (FS#23378). I also cleaned up the redirection.
Signed-off-by: Jan Steffens<jan.steffens@gmail.com> Fine with the patch, but can I ask how rigorous the testing is this has gone through? I'd want to be relatively certain we don't regress here since this will land on maint.
I have tested this against all situations I can remember that the original logpipe stuff was supposed to have fixed. Unless I am missing something, this is good to go with the minor change I show below. Version with that change is on my working-maint branch.
--- scripts/makepkg.sh.in | 11 +++++------ 1 files changed, 5 insertions(+), 6 deletions(-)
diff --git a/scripts/makepkg.sh.in b/scripts/makepkg.sh.in index 5742c49..0b3891e 100644 --- a/scripts/makepkg.sh.in +++ b/scripts/makepkg.sh.in @@ -794,15 +794,14 @@ run_function() { # ensure overridden package variables survive tee with split packages logpipe=$(mktemp -u "$startdir/logpipe.XXXXXXXX") mkfifo "$logpipe" - exec 3>&1 - tee "$BUILDLOG"< "$logpipe"& - exec 1>"$logpipe" 2>"$logpipe" + tee "$BUILDLOG"< "$logpipe"& teepid=$!
put these on separate lines and make teepid local. tee "$BUILDLOG" < "$logpipe" & local teepid=$!
restoretrap=$(trap -p ERR) trap 'error_function $pkgfunc' ERR - $pkgfunc 2>&1 + $pkgfunc&>"$logpipe" eval $restoretrap - sync - exec 1>&3 2>&3 3>&- + + wait $teepid rm "$logpipe" else restoretrap=$(trap -p ERR) -- 1.7.4.1