\Ucharcat is available in recent (u)pTeX

It can only cover the 8-bit range, but that's OK
as there is a check before we ever try to call it.

l3kernel/l3token.dtx

@@ -1557,12 +1557,11 @@
   }
 \tl_new:N \l_@@_tmp_tl
 %    \end{macrocode}
-%   Engine-dependent definitions are now needed for the implementation. For
-%   \LuaTeX{} and \XeTeX{} there is engine-level support.
-%   They can do cases that macro emulation can't. All of those are filtered
-%   out here using a primitive-based boolean expression to avoid fixing
-%   the category code of the null character used in the false branch
-%   (for 8-bit engines).
+%   Engine-dependent definitions are now needed for the implementation. Recent
+%   (u)p\TeX{} and the Unicode engines \LuaTeX{} and \XeTeX{} have engine-level
+%   support for expandable character creation. \pdfTeX{} and older (u)p\TeX{}
+%   releases do not. The branching here if low-level to avoid fixing
+%   the category code of the null character used in the false branch.
 %   The final level is the basic definition at the engine level: the arguments
 %   here are integers so there is no need to worry about them too much.
 %   Older versions of  \XeTeX{} cannot generate active characters so we filter
@@ -1573,9 +1572,7 @@
 \group_begin:
   \char_set_catcode_active:N \^^L
   \cs_set:Npn ^^L { }
-  \if_int_odd:w 0
-      \sys_if_engine_luatex:T { 1 }
-      \sys_if_engine_xetex:T { 1 } \exp_stop_f:
+  \if_cs_exist:N \tex_Ucharcat:D
       \cs_new:Npn \@@_generate_aux:nnw #1#2#3 \exp_end:
         {
           #3