Certain string.format operations produce clearly wrong results, for
instance "string.format("%d", 2^63)" gives -9223372036854775808 on
amd64 yet 9223372036854775807 on riscv; This is the same as with 5.2 but
unlike the latter, it doesn't trigger a test failure here due to
much smaller test coverage of 5.1. However:
- an upstream mailing-list discussion from 2011 [1] suggests
string.format("%d") is known to be borked in Lua 5.1 for very large
integers;
- this is unlikely to have an effect on arithmetics because Lua numbers
are floating-point (string formatting this is one of the few places
in the interpreter where they get internally converted to integers)
Not to mention that this doesn't work correctly on several other arches
for which lua:5.1 has already not only been keyworded but actually
stabilised.
[1] http://lua-users.org/lists/lua-l/2011-07/msg00876.html
Signed-off-by: Marek Szuba <marecki@gentoo.org>
Marek Szuba
082ff82e32