On Sun, 7 Sep 2008 [EMAIL PROTECTED] wrote:
Author: jmb
Date: Sun Sep 7 08:32:01 2008
New Revision: 5270
URL: http://source.netsurf-browser.org?rev=5270&view=rev
Log:
Native RO makefile. Don't expect anything other than make/make debug to
actually work thanks to insane amounts of buggy behaviour in the
OS/runtime that I can't find the energy to work around.
For those who actually care about this, here's what does/doesn't work:
Works:
make
make debug (though no debug symbols, for obvious reasons)
Does not work:
make clean
+ rm -rf build/Release fails inexplicably. rm -rf build works,
except that it fails to delete .svn. This looks like a bug in rm
to me.
make distclean
+ as above
make coverage
+ no lcov on RO
make {profile,export,install,uninstall}
+ I've not actually tested these, but would be amazed if they
worked.
make test
+ Dependency tracking on the debug library fails to work, so I've
hacked around it such that the test binaries actually build.
(Make fails to work out that
<LibParserUtils$Dir>/libparserutils-debug.a actually
exists, or not, as the case may be, then complains that there's
no rule to make the thing)
+ The testrunner utterly fails thanks to what looks like broken
stream redirection (stdout, and probably stderr from the child
task never gets sent to the stream handles created for them in
the parent perl script)
Oh yeah, there's also the hideous hackery in Makefile-ronative that's
caused by Make objecting violently to colons in paths.
I have no intention of expending any more effort on this. If someone
actually wants to use this buildsystem on RO, then they get to fix it.
The same will apply to hubbub's buildsystem when I make the same changes
to it.
J.
