summaryrefslogtreecommitdiff
path: root/SConstruct
AgeCommit message (Collapse)AuthorFilesLines
2010-05-28Experimentally add support for a suite of tests based on gtest, and convert ↵Daniel Burrows1-2/+7
the logging tests to it. gtest has some nice features for producing extra information in failures that could be handy. Also, it integrates well with gmock; the gmock failures in my Boost tests were sometimes kind of obscure.
2010-04-29Fix how the main source list is generated.Daniel Burrows1-1/+2
Since I want sources (for use with the .po rules), I should call srcnode() on each file. This also means the sanity-check I put in to catch this sort of thing doesn't bail.
2010-04-27Ensure that the master source list in the top-level file is always in the ↵Daniel Burrows1-0/+3
same order. Otherwise scons will rebuild stuff that depends on this (like the .pot file) when the order changes.
2010-04-26Add an SConscript file for the po/ directory.Daniel Burrows1-0/+15
2010-04-26Flatten the list of aptitude sources before making a set from it, since only ↵Daniel Burrows1-1/+1
the leaf nodes should be stored. Otherwise we try to put lists into a set, which is just wrong.
2010-04-26Return the list of all aptitude's sources from the top-level SConscript ↵Daniel Burrows1-1/+11
file, so that it can be used to generate pofiles. This is a bit ugly IMO, and I seriously considered the alternative of writing a list of *all* the source, putting it in a module, and importing that into both src/SConscript and SConstruct. However, there are lots of source files in aptitude, and splitting the build files up would be moderately annoying.
2010-04-25Build the documentation in a variant directory again.Daniel Burrows1-4/+1
Hopefully this won't break anything...
2010-04-25Disable the use of an alternate tree for the documentation, to reduce the ↵Daniel Burrows1-1/+4
number of variables involved in its breakage.
2010-04-25Don't try to duplicate the documentation source files.Daniel Burrows1-1/+1
In theory, duplicating the source files makes scons work better. In practice, scons randomly "forgets" to copy files when it needs them, then complains and bails out of the build because they aren't present because it didn't copy them.
2010-04-25Write a build skeleton for the docs directory that's complete enough to ↵Daniel Burrows1-0/+9
build the Czech documentation.
2010-04-25Use the new Install builder to write installation rules for aptitude's ↵Daniel Burrows1-0/+26
executables and runtime data.
2010-04-25Distribute the rest of the stuff that the toplevel Makefile.am would distribute.Daniel Burrows1-1/+37
2010-04-25Explicitly distribute the build files in SCons*.Daniel Burrows1-1/+6
This seems to be the least worst option I can see. Since my Dist() builder is outside scons and scons doesn't seem to offer any hooks or listeners, I can't (safely) cause build files to always be added to the distribution list. And anyway, I'm not sure I want to. Better to make the distribution fully explicit.
2010-04-25Start adding support for a 'dist' rule that builds a distributed archive.Daniel Burrows1-0/+20
Before this can work, I need to add all the SConscript files to the archive explicitly, which is yucky.
2010-04-24Build the unit test outputs in the appropriate variant directories, not the ↵Daniel Burrows1-1/+3
last directory containing program output.
2010-04-24Change build/.../test to build/.../tests to match the source tree.Daniel Burrows1-1/+1
2010-04-24Use the new test builder to make sure the unit tests build properly and get ↵Daniel Burrows1-3/+3
executed.
2010-04-24Make it easier to get relative paths between tests and sources.Daniel Burrows1-2/+2
2010-04-24Move the pkg-config checks into the configure code.Daniel Burrows1-13/+0
Not only do they get cached this way, they'll be available for the child environments.
2010-04-24Try to build and run unit tests, and only build aptitude by default.Daniel Burrows1-1/+11
TODO: probably should add a UnitTest builder.
2010-04-24Move more setup code into aptitude_configure.py.Daniel Burrows1-44/+3
This step also creates a unit test build environment. TODO: if unit test configuration fails, it should just not be possible to compile the unit tests (instead of failing the build), as a nicety to people who "just want to compile".
2010-04-23Break up the configure-related code into its own Python files.Daniel Burrows1-26/+3
This creates a file for the configure code itself (from SContruct), a file for the custom checks the configure code uses, and a file for other general code related to the configure process.
2010-04-23Add canned compile flags for profiling builds.Daniel Burrows1-1/+2
2010-04-22Merge with desktop.Daniel Burrows1-5/+16
2010-04-22Use the new variant infrastructure to support variant builds.Daniel Burrows1-5/+3
2010-04-22Add a simple scheme for defining variants.Daniel Burrows1-0/+11
The idea is that the build script defines several "axes" of configuration, with one "option" on each axis. A variant selects one option along each axis (the terminology is a bit confused and I sometimes call options variants). The specification of what to build is flexible, which was my main goal: If you have three axes: Interface: curses, gtk Build options: debug, optimized Build flavor: orange, lemon you get these variants: [curses-debug-orange, curses-debug-lemon, ...] and you can select them like this: all # All the variants curses-orange # All variants that choose curses AND orange lemon-optimized # All variants that choose lemon AND optimized debug # All debug builds curses,lemon-gtk # Anything that's curses, plus lemon-gtk curses,debug # Anything that's curses OR debug Each variant uses a separate build directory (build/variant-name). I don't expect that aptitude will have more than 2-3 axes with 3-4 options per axis, so this should stay reasonable.
2010-04-22Add a configure test for boost IOStreams.Daniel Burrows1-0/+2
2010-04-22Merge with laptop.Daniel Burrows1-0/+8
2010-04-21Add two more directory definitions.Daniel Burrows1-0/+6
Actually, lockfile is a file. Really, there isn't anything directory-specific about DefineDirectory; maybe its name should be changed?
2010-04-21Define HAVE_DDTP if apt appears to include the DDTP patch.Daniel Burrows1-0/+2
2010-04-21Bump the version number to 0.6.2.1.Daniel Burrows1-1/+1
2010-04-21Use the DefineDirectory tool to clean up directory definitions in SConstruct.Daniel Burrows1-69/+21
2010-04-21Add more directory options.Daniel Burrows1-4/+30
This is probably almost it. I want to make all of these consistent, maybe using a function to generate them; I just need to figure out the best way.
2010-04-21Fix the definition of the various "dir" directories to recursively expand them.Daniel Burrows1-3/+3
2010-04-21Add more stock directories.Daniel Burrows1-3/+28
2010-04-21Define LOCALEDIR and PREFIX to their usual values in SConstruct.Daniel Burrows1-2/+20
2010-04-21Define SIGC_VERSION the same way that the configure script does.Daniel Burrows1-0/+3
2010-04-21Define ENABLE_NLS to 1, not "".Daniel Burrows1-2/+1
Some bits of the aptitude code do "#if ENABLE_NLS", not #ifdef.
2010-04-21Define PACKAGE and VERSION preprocessor symbols, and enable config.h generation.Daniel Burrows1-0/+5
2010-04-21Improve the gettext detection to generate #defines appropriately.Daniel Burrows1-1/+1
Also generally cleaned it up.
2010-04-21Add a skeletal routine to find gettext and friends.Daniel Burrows1-0/+3
Needs to be beefed up a little.
2010-04-21Don't print the environment at the end of the build script.Daniel Burrows1-2/+0
2010-04-21Add a test for the presence of PThreads.Daniel Burrows1-0/+2
2010-04-21Don't run the C++ compiler check.Daniel Burrows1-1/+0
The reason is purely aesthetic: it isn't generally necessary, and its status messages stuck. They look like this: Checking for a working C++ compileryes
2010-04-21Add a custom test for PO4A.Daniel Burrows1-0/+1
Doesn't use the @ConfigureCheck decorator since I don't want that behavior.
2010-04-21Fix a missing import.Daniel Burrows1-0/+1
2010-04-21Fail to configure if a required component can't be found.Daniel Burrows1-5/+6
2010-04-21For extra paranoia, check that the C++ compiler actually works.Daniel Burrows1-0/+1
2010-04-21Always put the top-level project directory and the top-level source ↵Daniel Burrows1-1/+2
directory into the include path.
2010-04-20Experiment with SCons a bit.Daniel Burrows1-0/+37
The SCons files I've added aren't enough to build aptitude yet.