result. The solution is different from the patch suggested in the PR, since
the debug message was only meant to be used during early development of
flavors support. In my local development version of portmaster, all output
from pm_v is redirected to STDERR, but that change seemed to disruptive to
apply to the "published" version.
Unrelated changes: make pm_isdir_pd more robust and fix variable name of
BACKUP_FORMAT option in sample file.
PR: 225843
Reported by: John Hein
Approved by: antoine (implicit)
The behavior of the ask-license target should be to silently succeed instead
of returning an error, IMHO, but I can test for this condition and skip the
ask-license phase in portmaster, too.
While here, I'm slightly simplifying the fix implemented in r460294.
PR: 225699
Submitted by: dewayne@heuristicsystems.com.au
Approved by: antoine (implicit)
a prefix of the package name with a new port to be installed. The cause of
this bug was that a pattern was applied without anchor at the end.
Testing revealed that "pkg info -x" ignores an anchor at the end of the
pattern (which might be a bug in pkg), therefore the output of the query
is now additionally filtered with egrep to obtain the desired result.
PR: 225496
Reported by: Martin Birgmeier
Approved by: antoine (implicit)
installed (e.g. for make targets "extract", "patch" or "stage" in build
dependencies).
Limit depth of search for empty directories to reduce the scan time if
there are non-empty work directory sub-trees.
Add package format parameters to rc.sample file.
Approved by: antoine (implicit)
an optional make target. The patch file contained 2 typoes ("{$" instead
of "${").
Add clean phase for dependencies that are not installed but required when
the dependent port is built (e.g. because of references to source files of
that port, or because of static linking against libraries in the stage area
of such a dependency).
Approved by: antoine (implicit)
This is required for e.g. net/openntpd, which links against static libressl
libraries in that port's staging area.
In case of a port that is not fully installed, there is no clean phase for
the dependency's port directory. The dependency tracking does not maintain
the necessary state to perform this cleanup when the depending port is
finished.
PR: 213887
Reported by: blackmore@pichove.org (Simeon Simeonov)
Approved by: antoine (implicit)
The patch is based on ideas presented in the PR, but implemented
differently. Re-open the PR, if this solution has unexpected side-effects
or proves insufficient.
PR: 174729
Reported by: ngie
Approved by: antoine (implicit)
version, if PYTHON3_DEFAULT is set to anything but 3.6. Since the MOVED
file sets the new origin and flavor such that a Python-3.6 compatible
package is built, the original Python version for this port is ignored.
Reported by: lev
Approved by: antoine (implicit)
circumvent a check for an empty FLAVOR parameter being passed to certain
ports.
Reported by: Walter Schwarzenfeld <w.schwarzenfeld@utanet.at>
Approved by: antoine (implicit)
before each commit and I'll whether I can create a pre-commit script that
checks this for me.
Reported by: Walter Schwarzenfeld <w.schwarzenfeld@utanet.at>
Approved by: antoine (implicit)
a useful feature, and the txz-format is currently hard-wired in a lot of
places in portmaster, but there has been a request to allow creation of
packages in a non-standard archive format.
Approved by: antoine (implicit)
string at build time had been replaced by a version string from a test
version.
Reported by: herbert@gojira.at (Herbert J. Skuhra)
Approved by: antoine (implicit)
backups with no or less compression than the default of "txz" provides.
Support for different archive format for packages in general could be
provided, but not added at this time, since I'm not convinced that this
would be a useful feature (please let me know, if you disagree ...).
Approved by: antoine (implicit)
names. The character '@' is illegal in such names and must be converted to
a legal character same as [-+./].
remobve post processing of "pkg check -dn" output, which was a left-over
from pre-PKG_NG times.
Reported by: koobs (invalid @), tatsuki_makino@hotmail.com (pkg check)
Approved by: antoine (implicit)
The patch attached to PR 181450 did no longer apply cleanly, therefore I
edited portmaster to follow the intent and in part the letter of the patch.
This patch should not affect port building in any way. I have no use for
the feature (and I'm still considering to completely remove support for the
installation of packages in portmaster), and therefore did not test that
it performs as expected by the submitter.
PR: 181450
Submitted by: rozhuk.im@gmail.com
Approved by: antoine (implicit)
origin and package name (sans version), as was the case for e.g. the
lang/cython3 port, which was moved to lang/cython@py36 with a package
name change from cython3-$version to py36-cython-$version.
Approved by: antoine (implicit)
being identified and replaced when the new package is installed.
Add library dependencies to the list of build dependencies to consider.
Reported by: Robert Huff
Approved by: antoine (implicit)
previous version's problem and I forgot to
Fix shbang: Remove the -x that had been added for debugging.
Reported by: (Walter Schwarzenfeld) w.schwarzenfeld@utanet.at
Approved by: antoine (implicit)
place than before flavor support was added. Move its definition to an
appropriate place.
Silence one pkq query command that may be called for a non-existing
package name.
Approved by: antoine (implicit)
Reported by: Milan Obuch, Paul Beard
supports upgrades from non-flavored port versions based on the information
in the MOVED file.
For initial installations of flavored ports, the flavor must be specified
as part of the port origin, e.g. "devel/py-py@py36" for the Python-3.6
version of that port. Dependent ports will automatically be installed with
the correct flavor passed via the dependency mechanism.
It is planned to add a --flavor option to ease flavor selection for ports
that are initially installed with portmaster.
Approved by: antoine (implicit)
the variable $unique_list was created with identical content, its value was
only set within the implicit sub-shell (2nd command in a pipe) and lost on
exit from the while loop.
Reported by: Jakub Laach
indefinite amounts of time, preventing installation or upgrade of
ports with portmaster. Instead of using a pipe to read pkg query
input to be consumed by a shell loop, buffer the query result in a
shell variable and let the loop iterate over the contents of this
variable.
While here, use a call to all_pkgs_by_origin() instead of an inline
versin of equivalent code.
Reviewed by: tz
Approved by: tz
Torsten has accepted the challenge of whipping portmaster back into
shape. In addition to changing MAINTAINER, he's starting with fixing
both non-default options by providing their necessary RUN_DEPENDS.
PR: 207075
