tests/Makefile | 5 4 + 1 - 0 !
1 file changed, 4 insertions(+), 1 deletion(-)

 hack to generate an empty config.h

tests/glib.lua | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 [patch] replace busy loop with usleep call

tests/glib.lua counts to one million in an attempt to spend some time so
that it can test that GLib's Timer correctly nootices that some
timepassed. The "count to one million"-loop needs about three
milliseconds here with Lua 5.2. LuaJIT is likely faster than this.

Instead of relying on this loop being slow enough, replace it with a
call to g_usleep to sleep for a millisecond.

Signed-off-by: Uli Schlachter <psychon@znc.in>
Debian-Bug: https://bugs.debian.org/880410

lgi/callable.c | 2 1 + 1 - 0 !
1 file changed, 1 insertion(+), 1 deletion(-)

 [patch] callable: fix enum return from c functions

According to [1], return values smaller than sizeof(long) are returned
as ffi_arg or ffi_sarg. Thus, the code here needs to use ffi_sarg
instead of int to work correctly.

This fixes a test failure on s390x that boils down to
cairo_pattern_get_type() returning 2, but the C code here turns that
into a value of 0.

A big "thank" you to Debian and specifically to Mattia Rizzolo for
giving me access to a porterbox so that I could debug this myself.

[1]: https://linux.die.net/man/3/ffi_call

Signed-off-by: Uli Schlachter <psychon@znc.in>
Bug-Debian: https://bugs.debian.org/879956

lgi/Makefile | 2 1 + 1 - 0 !
tests/Makefile | 2 1 + 1 - 0 !
2 files changed, 2 insertions(+), 2 deletions(-)

 honour cppflags

tests/pango.lua | 3 1 + 2 - 0 !
1 file changed, 1 insertion(+), 2 deletions(-)

 adapt to new pango 1.56.2 bindings