diff --git a/core/shared/src/main/scala/fs2/concurrent/TimedPull.scala b/core/shared/src/main/scala/fs2/concurrent/TimedPull.scala
index eec57e3e56..17bd38fa11 100644
--- a/core/shared/src/main/scala/fs2/concurrent/TimedPull.scala
+++ b/core/shared/src/main/scala/fs2/concurrent/TimedPull.scala
@@ -35,9 +35,6 @@ object tp {
     def startTimer(t: FiniteDuration): Pull[F, INothing, Unit]
   }
   object TimedPull {
-    // newTimeout
-    // change Token in TimedPull
-    // move filter of stale timeouts in the Pull
     def go[F[_]: Temporal, A, B](pull: TimedPull[F, A] => Pull[F, B, Unit]): Pipe[F, A, B] = { source =>
       def now = Temporal[F].monotonic
 
@@ -52,7 +49,7 @@ object tp {
         def nextAfter(t: Timeout): Stream[F, Timeout] =
           time.discrete.unNone.dropWhile(_.id == t.id).head
 
-        // is the initial time.get.unNone fine or does it spin?
+        // TODO is the initial time.get.unNone fine or does it spin?
         def timeouts: Stream[F, Token] =
           Stream.eval(time.get).unNone.flatMap { timeout =>
             Stream.eval(timeout.asOfNow).flatMap { t =>
@@ -65,7 +62,8 @@ object tp {
           timeouts
             .map(_.asLeft)
             .mergeHaltR(source.chunks.map(_.asRight))
-            .evalMapFilter {
+            .evalMapFilter { // TODO filter of stale timeouts in uncons
+
               case Right(c) => c.asRight[Token].some.pure[F]
               case Left(id) => time.get.map(t => t.filter(_.id != id).as(id.asLeft[Chunk[A]]))
             }
@@ -97,6 +95,7 @@ object tp {
         if (c.size < n) s -> c
         else {
           val (unit, rest) = c.splitAt(n)
+          // TODO use Chunq.queue here instead of aggregating via emission?
           resize(rest, s >> Pull.output1(unit))
         }