| Title | Hash table advice does not mention sets |
| Status | closed |
| Priority | nice |
| Assigned user | Gareth Rees |
| Organization | Ravenbrook |
| Description | The documentation for address-based hash tables [1] explains that you have to test for staleness when setting a value for a key in a hash table. However, if the hash table is being used to implement a set, and thus there are no values, only keys, then you can skip this step -- in a set it is fine to have a key present multiple times (because there is no need to figure out which instance is the correct one), so long as the rehash mechanism can cope with this. |
| Analysis | Document this but explain that it is an optimization. |
| How found | inspection |
| Evidence | [1] https://www.ravenbrook.com/project/mps...guide/advanced.html#location-dependency |
| Created by | Gareth Rees |
| Created on | 2018-09-19 13:21:08 |
| Last modified by | Gareth Rees |
| Last modified on | 2019-01-29 15:42:11 |
| History | 2018-09-19 GDR Created. |
| Change | Effect | Date | User | Description |
|---|---|---|---|---|
| 195953 | closed | 2019-01-29 15:42:11 | Gareth Rees | Add note about skipping the staleness check for sets. |