Mercurial > pylearn
comparison doc/v2_planning/requirements.txt @ 1205:5525cf3faaa2
requirements: Question about the serialization requirement
| author | Olivier Delalleau <delallea@iro> |
|---|---|
| date | Tue, 21 Sep 2010 10:48:01 -0400 |
| parents | ab80ba052d32 |
| children | 31b72defb680 |
comparison
equal
deleted
inserted
replaced
| 1204:d9ffb9ebf8ac | 1205:5525cf3faaa2 |
|---|---|
| 110 at deserialization time. | 110 at deserialization time. |
| 111 | 111 |
| 112 This requirement is aimed at enabling process migration and job control as well | 112 This requirement is aimed at enabling process migration and job control as well |
| 113 as post-hoc analysis of experiment results. | 113 as post-hoc analysis of experiment results. |
| 114 | 114 |
| 115 OD asks: When you say "The experiment cannot assume that resources written by | |
| 116 the experiment will still be there", do you mean we should be able to recover | |
| 117 the exact same output after interrupting an experiment, wiping its expdir, and | |
| 118 restarting it? This would mean that any output saved on disk by the experiment | |
| 119 also has to be serialized within the experiment, which may lead to very big | |
| 120 serialization files (and possibly memory issues?) | |
| 121 A less constraining interpretation of your statement (which I like better) is | |
| 122 that we allow "previous" output to be lost: we only ask that the experiment | |
| 123 should be able to produce the "new" outputs after a wipe+restart. |