Clamping of consecutive elements with a common parent is broken

[codener]

Consider this markup:

<div class="common-parent">
  <div class="first" clamp>Some ... text</div>
  <div class="second" clamp>Some ... text</div>
</div>

In this situation, Superclamp will not always clamp the .second element properly, i.e. its contents will exceed its height and bleed out of .second. Unfortunately, this behavior is not deterministic, meaning it will sometimes clamp right and sometims do it wrong.

When investigating a broken .second, I found that Superclamp had stored incorrect elementAt values for the element.

I assume this originates from the fact that Superclamp splits "measurement" and "adaption" in two consecutive phases. This breaks when the measures for .second are changed during the adaption of .first.

[makmic]

I managed to create a minimal example, where the second of two subsequent elements is not clamped correctly.

c646b6c

As this error only occurs when using max-height, my naive guess would be to process clamping jobs ordered via LIFO, that the second element is always processed before the first. (Code)

drainPhaseQueue = (phase) ->
  jobs = jobQueues[phase]
  if jobs.length > 0
-    while job = jobs.shift()
+    while job = jobs.pop()
      job()

Sadly, this does not fix the issue. Any other ideas?

[makmic]

One more hint: Disabling value-caching fixes the problem, but is not an option:

- instance = element[INSTANCE_KEY] || new Superclamp(element)
+ instance = new Superclamp(element)

[foobear]

Fixed in 0.2.3