Eric Blake <ebl...@redhat.com> writes: > On 05/04/2016 10:49 AM, Markus Armbruster wrote: >> Type QJSON lets you build JSON text. Its interface mirrors (a subset >> of) abstract JSON syntax. >> >> QAPI output visitors also produce JSON text. They assert their >> preconditions and invariants, and therefore abort on incorrect use. >> >> Contrastingly, QJSON does *not* detect incorrect use. It happily >> produces invalid JSON then. This is what migration wants. >> >> QJSON was designed for migration, and migration is its only user. > > Worth calling out commits 0457d07..b174257 here?
Can't hurt, will do if I need to respin. >> Move it to migration/ for proper coverage by MAINTAINERS, and to deter >> accidental use outside migration. >> > >> +++ b/include/migration/vmstate.h >> @@ -29,7 +29,7 @@ >> #ifndef CONFIG_USER_ONLY >> #include <migration/qemu-file.h> >> #endif >> -#include <qjson.h> >> +#include "migration/qjson.h" > > I thought you weren't a fan of including .h from .h, where it was > avoidable. But I guess you aren't adding any new .h, so much as > converting an existing use. Correct. >> + >> #include "qemu/osdep.h" >> -#include <qapi/qmp/qstring.h> >> -#include <glib.h> >> -#include <qjson.h> >> -#include <qemu/module.h> >> -#include <qom/object.h> >> +#include "qapi/qmp/qstring.h" >> +#include "migration/qjson.h" >> +#include "qemu/module.h" >> +#include "qom/object.h" > > Thanks for fixing the mis-use of <> while at it :) I intend to do that globally, in my header cleanup project. >> +++ b/migration/vmstate.c >> @@ -6,7 +6,6 @@ >> #include "qemu/bitops.h" >> #include "qemu/error-report.h" >> #include "trace.h" >> -#include "qjson.h" > > This is because you are relying on the .h doing it for you. Yes. > As mentioned on the cover letter, > Reviewed-by: Eric Blake <ebl...@redhat.com> > whether or not you touch up the commit message to call out ids Thanks!
