* new/project.jam
(inherit-attribute): New rule, extracted from 'initialize'.
(initialize): Allow 'jamfile' parameter to be empty, in which case
the project is 'standalone'.
[SVN r18544]
they are located. The problem with using the directory name is that we might
want toolset modules to act as project, and directory name is not unique then.
We might even want to declare two projects in the same module.
[SVN r18542]
- darwin-tools.jam; added warnings feature to control warning output.
- darwin-tools.jam; disable some Apple speicfic options when using g++ directly.
- darwin-tools.jam; disable dylib versioning, for now.
- python.jam; remove shared link to pyton2.3 lib for embebed targets.
- python.jam; remove uneeded path to python libraries, implicitly there because of framework use.
- python.jam; disable warnings, for now.
[SVN r18529]
- darwin-tools.jam; added GCC* env setup variables to allow for using something other than the built in gcc.
- darwin-tools.jam; tweaked the bundle-loader feature to also add the bundle as a link object.
- python.jam; wire in the framework path for python.
- python.jam; remove the <framework> feature from built PYDs.
[SVN r18522]
* new/stage.jam
(stage-target-class.construct): Pass the result via
'virtual-target.register'. I wonder if virtual targets should
be create via 'virtual-target.create' which will invoke
'virtual-target.register' internally. Passing via 'register' was forgotten
in many places.
* test/stage.py: New test.
[SVN r18497]
* new/make.jam
(make): Pass 'default-build' in the right position.
* new/targets.jam
(basic-target): Use empty property-set if no default-build is passed.
(main-target): Use 'raw' to test property-set for emptines. The string
emptyness test that use to be there does not works for property-sets.
[SVN r18440]
default build was stored as plain list, and implied properties were not
converted. E.g. we had "debug release", not "<variant>debug <variant>release",
which was assumed.
* new/targets.jam
(main-target, basic-target): Store 'property-set' as default-build
attribute.
(main-target-default-build): Use 'property-set.create-with-validation'.
* new/property-set.jam
(create-with-validation): New rule.
* test/default_build.py: More tests.
[SVN r18436]
translation. Cleaned up indentation and idiomatic use of string.
* index.html - documented recent features including VARNAMES and
DELETE_MODULE
* borland.jam, msvc.jam - use the new :W modifier to simplify
code (but only somewhat)
* BoostBuild.py - ignore MSVC .ilk files under Cygwin and NT
[SVN r18434]
Consequently rolled back the workaround in make_rule.py
* Added new indirect rule invocation module for encoding the module to
invoke a rule from as well as the exact name to invoke it by, and
for subsequent invocation.
* Fixed module __test__ rules so they really do execute in a separate
module. Associated tweaks to __test__ in path.jam
[SVN r18427]
* new/virtual-target.jam
(null-action): Pass '.no-action' as action name, not 'no-action'. The
action name is really dummy, it's not used at all. However, when it does
not include dot, the smart 'action' ctor will assume it's local and prepend
module name to it. The calling module will be 'null-action' instance module,
always unique, and as result, no two targets with null actions will be
equal, as far as 'virtual-target.register' is concerned.
[SVN r18424]
of all actions that jam could possibly execute is not good-looking. E.g:
gcc.compile
gcc.link-dll
gcc.compile
gcc.link
...found 19 targets...
...updating 9 targets...
MkDir1 app/bin/gcc
MkDir1 app/bin/gcc/debug
MkDir1 app/bin/gcc/debug/main-target-app
compile app/bin/gcc/debug/main-target-app/app.o
[SVN r18418]
* handle the absence of QT more gracefully
* flush stdout after each test so that emacs shows progress
****** Commented out tests which appear to be failing outside my control ******
[SVN r18410]