job_pipe and collect another one for the next job.
If we are aborting, remove all the 'normal' job tokens and add an 'error' one.
If we get an 'error' token, remove any other tokens, re-insert the error
token and exit (with error 'cos that is easier).
Add the current pid to some of the DEBUG(JOB) traces.
Combined effect is that parallel makes actually stop some fairly shortly
after an error, rather than running on long enough to fill the scrollback.
output in parallel makes.
After all with -s you wouldn't know the command for a non-parallel make.
Makes (sic) the output of parallel NetBSD build fathomable.
Lpd has the ability to start on a port other than "printer" but has no
way to connect to a remote lpd on that port. This change adds the
ability to specify port@host in the rm element of printcap(5). Tested to
work with both a standard lpd setup and one running on special ports.
specification in .../mtree/NetBSD.dist, run ${TOOL_MTREE} again with
different args to register the new directories in the metalog.
Previously, these directories were not mentioned at all
in the metalog.
Reviewed by agc
to the existing check for consistency between DESTDIR and the output
from makeflist.
* Use egrep instead of awk to ignore differences that are expected.
This should be easier to maintain, and should also make it easier
for users to add their own custom exceptions by editing the
regexp.
Reviewed by agc
"makesyspkgsums" targets for it to use internally.
* Add internal "sanitize_METALOG" target that uses "${TOOL_MTREE} -C ..."
to sanitise the METALOG file before invoking anything that will
want to parse the METALOG (checkflist, regpkgset, maketars, or
installsets). Previously, maketars used to do this itself, but its
better to do it only once.
* Add internal "checkflist_if_DESTDIR" target, which invokes checkflist iff
DESTDIR is set and not empty. Use it where appropriate.
* Add a few more host tools to SETSENV.
* A few minor related changes.
Reviewed by agc
* Use dirname, don't try to use a shortcut that doesn't always work.
This uses the cheap implementation of dirname from sets.subr.
* Fix error in getopts string ("-p" option takes an arg).
* Explicitly use ${HOST_SH} to run culldeps script, because we might
be on a host where /bin/sh is not a POSIX shell.
Reviewed by agc
The new "-t binpkgdir" option requests this action.
* Make it pay attention to DESTDIR.
* Make it work for unprivileged builds using METALOG.
* Add "force" and "update" modes.
* Add "quiet" mode. There was already a "verbose" mode.
* Add several new command line args in support of the above.
* Make much more use of shell functions.
* Replace the old way of choosing syspkg version numbers.
The new way gives numbers derived from concatenating the OS
version [from osrelease.sh or $(uname -r)], a "tiny" version [from
distrib/sets/versions, default 0], and a date [from RCS time stamps or
file system time stamps].
* Add @blddep lines to the PLIST (in addition to the @pkgdep lines
that were previously added).
* Use host tools such as pax, cksum, and db, to do more or less the
same work that was previously done by pkg_create (which is not a host
tool). No longer attempt to use pkg_create.
Reviewed by agc
"-P ${DESTDIR} -I /". ("-I" chosen for similarity to pkg_create.)
XXX: makeplist seems to be quite broken, in that it looks at the files
and directories that are actually present in DESTDIR, whereas I
thought its job was to provide an list of what *should* be there,
regardless of what is *actually* there. Fixing this seems to require
a change to the format of the "lists" files, so that they can
unambiguously identify directories.
Reviewed by agc
locations:
- Don't declare pthread__switch_away global
- Do the PIC dance for pthread__switch_return_point and
pthread__locked_switch. Ideally these (and other) symbols would
be hidden.
Thanks to uwe@, dyoung@ and elad@ for help.
XXX sh3 is still to be done.
XXX vax does strange things.
Commit fixes problem reported (privately) by Jeff Rizzo.
XXX: Note that we are still are unable to allow a reconstruct of
a failed spare -- it's easy enough to attempt to do it, but the
code in rf_ContinueReconstructFailedDisk() isn't anywhere close to
ready to allow it. Yet another reason why the whole disk/spare thing
needs to be re-worked. *sigh*