Not sure if I understand. Do you have a demo? When resolving percentage widths on a table cell, it's pretty much irrelevant, since a row is exactly as wide as the table it's in. For heights, I'd say we do resolve them against the table height:

<table style="height:400px; background:pink;">
  <tr>
    <td style="height:80%; background:cyan;">~320px tall</td>
  </tr>
  <tr>
    <td>~80px tall</td>
  </tr>
</table>

The cell height is resolved against the used height of the row. Which I agree is special-casing for tables, and has to be done regardless of what the cell think its containing block is.

The CL for postion:sticky prompted me to open this bug because I remembered encoutering this underspecification before, probably when the resolution of percentage cell height was less well defined than it is now in CSS3.

If I ever think of something else that speccing this is relevant to I will add it here.

Oh and you're right - the table is the cell's containing block in Blink, I had it the wrong way round. I'm not sure it's obvious that's what it should be when you look at the spec though.