]> code.ossystems Code Review - openembedded-core.git/commit
db: change types to avoid headers changing between architectures
authorRoss Burton <ross.burton@intel.com>
Wed, 1 Nov 2017 13:52:19 +0000 (13:52 +0000)
committerRichard Purdie <richard.purdie@linuxfoundation.org>
Wed, 8 Nov 2017 22:23:43 +0000 (22:23 +0000)
commit60aa20b8b691e5a72a6a11bf795b86c6359db886
tree95cf195d174c2afe21b16e3f2464bccfcea86793
parent090eeccce74554bd4282b6a0407963037bc761a9
db: change types to avoid headers changing between architectures

Triggered by looking at why Python doesn't find db.h (because it greps db.h for
a regex, and db.h is actually a oe_multilib_header wrapper) I realised that the
only reason we have to oe_multilib_header db.h is because one typedef is
different between 32-bit and 64-bit architectures.

However, the typedef is for a 64-bit integer so instead of using long (64-bit)
or long long (32-bit), just use int64_t.  Some of the overly complicated
configure tests need to be deleted after this change but that is safe as we're
building in a controlled environment and can assume int64_t exists.

With this done the header doesn't change between architectures, and it doesn't
need to be wrapped by oe_multilib_header.

Signed-off-by: Ross Burton <ross.burton@intel.com>
meta/recipes-support/db/db/sequence-type.patch [new file with mode: 0644]
meta/recipes-support/db/db_5.3.28.bb