14 авг. 2019 г., в 14:12, Vladimir Davydov <vdavydov.dev@gmail.com> написал(а):

On Thu, Aug 08, 2019 at 02:55:54PM +0300, Serge Petrenko wrote:

---
src/lua/pickle.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/src/lua/pickle.c b/src/lua/pickle.c
index e47ac11b4..65208b5b3 100644
--- a/src/lua/pickle.c
+++ b/src/lua/pickle.c
@@ -109,14 +109,14 @@ lbox_pack(struct lua_State *L)
case 'I':
case 'i':
/* signed and unsigned 32-bit integers */
- if (field.type != MP_UINT && field.ival != MP_INT)
+ if (field.type != MP_UINT && field.type != MP_INT)
luaL_error(L, "pickle.pack: expected 32-bit int");

luaL_region_dup(L, buf, &field.ival, sizeof(uint32_t));
break;
case 'N':
/* signed and unsigned 32-bit big endian integers */
- if (field.type != MP_UINT && field.ival != MP_INT)
+ if (field.type != MP_UINT && field.type != MP_INT)
luaL_error(L, "pickle.pack: expected 32-bit int");

field.ival = htonl(field.ival);

I assume this is a bug and so this patch should be pushed to all
maintained branches, right?

Could you please add a test for this issue?