On 17 May 2019, Pierre JUHEN verbalised: > HI again, > >> I looked a bit on the code (maybe an old version), however, I saw in the macro defined by btree.c definition of variables of int type. > > Could it be a change in GCC, where the default int move from short to long and did have side effects ? That would be a massive ABI break and would break more or less all software compiled with GCC 9. So... definitely not, I'd say. :) -- NULL && (void)