Browse token
For Getnext requests on secondary indexes, a browse token is used to hold the name of the previously found entry. The token consists of the name found in the secondary index (which may not be unique) and the name in the primary index (which is unique).
The address of the directory entry cannot be used instead of this logical name because the entry may be returned unlocked, and so may be deleted when the next getnext request is received.
The getnext consists of locating the entry in the secondary index which has a the correct primary index, if it exists, and then moving forward in the getnext chain. If it does not, an entry with a matching secondary index name, but a higher primary index name is located, if one exists. If that also does not exist, an entry with a higher name in the secondary index is located. This requires that entries on the getnext chain for ordered both by secondary index name and also when identical secondary index names exist, by primary index name.