On Sun, Jan 9, 2011 at 10:11 PM, Lennart Augustsson <span dir="ltr">&lt;<a href="mailto:lennart@augustsson.net">lennart@augustsson.net</a>&gt;</span> wrote:<br><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex;">
That looks like it looses the efficiency of the underlying representation.<br></blockquote><div><br></div><div>Yes, I don&#39;t think one can retain that cleanly without using restricted monads to exclude things like</div>
<div><br></div><div>    liftM ($42) (mplus (return pred) (return succ))</div><div><br></div><div>or just</div><div><br></div><div>    liftM ($42) (return pred)</div><div><br></div><div>Maybe one can hack something to achieve</div>
<div><br></div><div>    mplus :: Ord a =&gt; Set a -&gt; Set a -&gt; Set a</div><div>    mplus a b = Set (\k -&gt; S.union (a &gt;&gt;- ret) (b &gt;&gt;- ret) `bind` k)</div><div>      where</div><div>        ret = S.singleton</div>
<div>        bind = flip Data.Foldable.foldMap</div><div><meta http-equiv="content-type" content="text/html; charset=utf-8"><div><br class="Apple-interchange-newline">    mplus :: not (Ord a) =&gt; Set a -&gt; Set a -&gt; Set a</div>
<div>    mplus a b = Set (\k -&gt; S.union (a &gt;&gt;- k) (b &gt;&gt;- k))</div></div><div><br></div><div>using overloading with undecidable instances (?) (and defining a Monoid instance for the Set monad in terms of the MonadPlus instance)</div>
<div><br></div><div>Reminds me of instance chains.. [1]</div><div><br></div><div>Sebastian</div><div><br></div><div>[1]: <a href="http://portal.acm.org/citation.cfm?id=1863596">http://portal.acm.org/citation.cfm?id=1863596</a></div>
</div>