| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
* Some packages rely on config files and such out of ${F} which are copied
into ${S} for some purpose. These files are not normally caught as a
dependancy. Now we can define these files in EXTRA_SOURCES so that builder
will know to rebuild a package if these files change.
|
|
* Add the builder git index and included configs to a packages source
dependancies.
|
|
* Add routine for including files found in ${BUILDER_CONFIG}
|
|
* The original idea behind the PROJECT variable is all but gone, and really ..
it wasn't very good to begin with. Remove any remaining references to it.
|
|
* We have pulled the environment setup from build-query and moved it into
load_rules(). This means that build-query acts to query information about a
package and is reserved more as a cmdline helper for external tools. Tools
that need package variables should use load_rules instead of
'build-query --environ'.
|
|
|
|
* Added a simple_usage() routine to builders defaults to help sub-commands
that do not have options (outside of --help).
|
|
* To aid in debugging we no longer delete stuff like the intermidiate
makefiles produced by Builder when invoked w/ 'build -d'
|
|
* We previously removed the SYSROOT declaration believing it to be a candidate
for the coming build-config command. It turns out that this not the case.
While SYSROOT can be set as a build-config directive, it is not optional for
builder to function.
|
|
|
|
* Added support for a sub-cmd to "import" the contents of another command.
This requires special rewrites to the sub-commands to be viable.
|
|
* We needed some way to detect when a sub-command was being called by the make
engine vs by a user so we could properly handle the 'help' command. This
involved a lot of decoupling of the make handling out of the top-level
'build' tool and placing it into a newly created 'build-make'.
* While we where at it we cleaned up the handling of various targets in
makedeps so that 'all/all' is now a valid target, and subsequently, the
default target.
|
|
* Cleaning up comments and variable uses in preparation for enabling
build-config
|
|
* Not really certain where it came from, and not really carrying .. it simply
doesn't belong.
|
|
* The reason for this option has been effectively lost to time, and is
effectively no longer necessary. The original reason it existed will be
replaced with a --config option in the future.
|
|
|
|
* Moved the test for gcc into dumpmachine directly, sort of curious that
dumpmachine doesn't take this sort of short-cut to begin with..
|
|
* Added a few new builtins to help performance in some minor use cases.
* Move generic builtin routines to top of build script
|
|
* New name better reflects the contents of this path.
|
|
* There are plans for a new routined named import() which has nothing to do
with loading of packages, or package rules.
* The new name better describes the process which is going on.
|
|
|
|
* Added a build-help sub-command. Trying to improve the quality of the help
environment so that each command can properly supply its own documentation.
|
|
* Detecting the number of CPU's is fairly OS specific, which is a bad road to
travel. More importantly though, MAKE_OPTS can simply be set in the environ
by the user for a given system, so automatically setting it gains little
outside of adding unecessary buzzword-compliance.
|
|
* While LICENSE is usually an optional package variable, it is required when
exporting packages to an RPM.
* In the future it may be worth-while to support exporting to other package
formats.
|
|
* In bringing builder versions in-line with other vendor tools (ebuild, rpm,
dpkg, etc) I am adding "optional" Release support.
* In order to "hopefully" integrate with existing Builder tree's the RELEASE
variable is allowed to be unset. If unset builder will attempt to derrive a
release from the VERSION variable. If no RELEASE can be found, then the
RELEASE defaults to 0.
|
|
* We use 'tee' to view our logs AND log it at the same time. Unfortunately
'tee' eats the exit status of 'make'. Bash has a solution for this via the
PIPEFAIL variable, but that approach is not entirely portablt. The only
solution I have been able to come up with that is reasonably portable is to
use a named pipe and put tee into the background. Fixes #52
|
|
* Incorrectly quoted the exit code when trying to fixup Issue #52, this fixes #57
|
|
* We use the 'tee' command to capture our output to the builder.log. The
problem is that 'tee' discards the exit status. Added a wrapper function to
capture the exit status and handed the output from the function to 'tee'.
* Fixes #52
|
|
* While dumpmachine and makedeps are technically valid builtin commands, they
are not valid "build" commands and can not be executed in any normal means.
Technically we should likely split out the valid top-level commands and the
library commands to avoid this sort of confusion.
* Fixes #13 @ 15m
|
|
|
|
|
|
* The 'build' command is now capable of acting as a script-wrapper for
sub-commands. This elininates the libbuilder.sh script and the fairly
verbose checking for libbuilder.sh in every sub-command.
|
|
* The Builder build engine is now working sufficiently enough to warrent
a proper SCM checkin. I have been adding enhanced functionality to
the system that demonstrates that I am approaching the end of the
proof-of-concept phase.
|
