if err != nil {
						err = msgp.WrapError(err, "Tiers", za0001)
						return
					}
					switch msgp.UnsafeString(field) {
					case "ts":
						za0002.TotalSize, err = dc.ReadUint64()
						if err != nil {
							err = msgp.WrapError(err, "Tiers", za0001, "TotalSize")
							return
						}
					case "nv":
						za0002.NumVersions, err = dc.ReadInt()
						if err != nil {

                }
                if ( ( next == null || chain.allowChain(next) ) && totalSize + size < maxSize && this.credits.tryAcquire(cost) ) {
                    totalSize += size;
                    last = chain;
                    chain = next;
                }
                else if ( last == null && totalSize + size > maxSize ) {

			}

			if sizeS.versions > 0 {
				res["versions"] = strconv.FormatUint(sizeS.versions, 10)
			}
			res["size"] = strconv.FormatInt(sizeS.totalSize, 10)
			for name, tier := range sizeS.tiers {
				res["tier-size-"+name] = strconv.FormatUint(tier.TotalSize, 10)
				res["tier-versions-"+name] = strconv.Itoa(tier.NumVersions)
			}
			if sizeS.failedCount > 0 {

  //     then total_size is the number of dimensions
  //     of the shape valued attribute, or -1
  //     if its rank is unknown.
  // (4) If type == TF_ATTR_SHAPE and is_list == 1
  //     then total_size is the cumulative number
  //     of dimensions of all shapes in the list.
  // (5) Otherwise, total_size is undefined.
  int64_t total_size;
} TF_AttrMetadata;

    std::vector<size_t> lens(expected.size());
    std::unique_ptr<char[]> storage(new char[m.total_size]);
    TF_OperationGetAttrStringList(op, tensorflow::kColocationAttrName,
                                  values.data(), lens.data(), expected.size(),
                                  storage.get(), m.total_size, s_);
    EXPECT_EQ(TF_OK, TF_GetCode(s_)) << TF_Message(s_);
    for (int i = 0; i < expected.size(); ++i) {