Re: [ANN] Lua 5.1.4-rc1 now available

[David Manura <dm.lua@...>]

Luiz Henrique de Figueiredo writes:
> Lua 5.1.4-rc1 is now available...

(1)

The docs on some luaL_* functions are not always precise concerning whether a
value must have a certain type or merely be convertible to that type.  For
example, luaL_checknumber has the description "Checks whether the function
argument narg is a number...", but it internally uses lua_tonumber/lua_isnumber,
which checks whether the value is merely convertible to a number.  Compare to
the documentation of lua_isnumber/lua_isstring/luaL_checkstring, which
explicitly states that conversion occurs.  The same might be said about a few
other functions, e.g. luaL_checkinteger, luaL_optnumber, etc.  However, this may
also equally apply to the documentation of many Lua functions like string.lower.
 string.lower(5) does work.  Maybe that is implied by "2.2.1 - Coercion":
"whenever a number is used where a string is expected, the number is converted
to a string, in a reasonable format".  2.2.1 does not say anything about
Booleans--e.g. every value is convertible to a Boolean, so lua_isboolean must
not do any coercion to be meaningful.

(2)

It would be nice to eliminate the _longjmp/_setjmp warning when compiling under
Cygwin using "make posix" (or "make linux"):

  [1] http://lua-users.org/lists/lua-l/2008-01/msg00367.html
  [2] http://lua-users.org/lists/lua-l/2006-01/msg00680.html
  [3] http://www.opengroup.org/onlinepubs/009695399/functions/_longjmp.html

It is caused by the LUA_USE_ULONGJMP define implied by LUA_USE_POSIX.  This
warning is harmless though, and I'm not sure how it affects other builds.