Added option to only run the compressor on a range of state groups in a room #44
Conversation
Azrenbeth
changed the title
This will later be used to mark whether a state group should be compressed or not
This will later get replaced with a min_state_group and a count of the number of state groups you wish to process up to.
Added function to database.rs that scans through room to find the id of the group in the room [groups_to_compress] after [min_state_group] This is printed out as part of the output (and the compressor can be continued by setting the min_state_group to this value). Also added some documentation to the config options in lib.rs
No longer reports each time it needs to find a missing predecessor
Azrenbeth
force-pushed
the
azren/added_chunking_option
branch
from
a39a398
to
9a12363
Compare
Why are we removing the option? I'd much rather we kept it and used a different name for the new option? That way we retain backwards compat. |
I removed it because I thought "max_state_group" was there just to limit the number of groups you work on at once, which the chunking options replace with something better, and removing it simplified the code a bit. And since there's no reason not to run the compressor on the state groups of the most recent messages (cos changing their deltas won't take THAT long) I couldn't think of a reason to keep it. (So mainly I just forgot to think about backwards compatability...!) Anyway, I'll add it back in! |
Co-authored-by: Erik Johnston <erikj@jki.re>
BUILDS ON azren/added_graph_option (PR #43)
This allows the compressor to be called on only a selection of state groups in a room.
-b can be used to select the state group number to start compressing from
(entering 0 or omitting this will start at the first state_group in the room)
-n can be used to enter (roughly) how many state groups you wish to compress.
The compressor will print to stdout where it got up to. Entering this state group number
on the next run of the compressor will start from where the previous run finished
(NOTE: chunking produces worse compression than running all in one go, but will save memory)