Will Robertson wrote:
 > If it's going to need such contents, would it make more sense to use a
> _toks datatype the whole way though?

Okay, I have "redone the exercise" in this way.

>> I wonder if there
>> is a reason not to do:
>>
>> \def_long_new:Npn \tlp_put_right:Nn #1#2{
>>  \tlp_set:Nx #1{\exp_not:o{#1}\exp_not:n{#2}}
>> }
>>
>> which does not suffer from the same issue.
> 
> Unless the whole point is to restrict its use in this case, I can't see
> why this isn't a better approach. I suspect it wasn't coded like this in
> the first place because \exp_not:n isn't available without eTeX. But I
> don't  think we should worry about that these days.

That was my guess too.
-- 
Joseph Wright