You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
In the original version of unique_objects_container, the hashes used to indicate uniqueness were written to file on saving and restored on loading. When it was found that this did not always yield the same hashes on different platforms, the hashes were instead regenerated on load.
The hashing algorithm has now changed and uses GetMD5 defined in the Horace code base. This means that the hashing should be independent of the platform used.
It is found that hashing can add significantly to load times, and when multiple small cuts are generated this slows down the use of Horace. If the hashes are added to the serialisation of unique_objects_container, the load should speed up.
Hence the issue is to add the hashes back into the serialization.
The text was updated successfully, but these errors were encountered:
In the original version of unique_objects_container, the hashes used to indicate uniqueness were written to file on saving and restored on loading. When it was found that this did not always yield the same hashes on different platforms, the hashes were instead regenerated on load.
The hashing algorithm has now changed and uses GetMD5 defined in the Horace code base. This means that the hashing should be independent of the platform used.
It is found that hashing can add significantly to load times, and when multiple small cuts are generated this slows down the use of Horace. If the hashes are added to the serialisation of unique_objects_container, the load should speed up.
Hence the issue is to add the hashes back into the serialization.
The text was updated successfully, but these errors were encountered: