HazyResearch / deepdive

DeepDive
deepdive.stanford.edu
1.96k stars 536 forks source link

"deepdive run" does less than "deepdive do all" #563

Open alldefector opened 8 years ago

alldefector commented 8 years ago

In my workflow, I often want/need to examine the weight table (dd_inference_result_weights_mapping), but "deepdive run" doesn't generate that. Instead, I have to run "deepdive do all" after "deepdive run" every time...

netj commented 8 years ago

deepdive redo init/app weights will do the job.

deepdive run was mainly kept for backward compatibility, but I guess we can refresh it to be something like: deepdive mark todo init/app && deepdive do all

xiaoling commented 8 years ago

@netj can you please confirm that init/app will not drop the database in any case?

I see deepdive db init in init/app/run.sh. I don't think this command is supposed to drop the database because db-init in both postgres and greenplum drivers doesn't dropdb unless the number of argument is greater than zero. But I did experience some cases when it did drop the db. Sorry can't recall the exact setting but was using the master branch a couple days ago.

netj commented 8 years ago

It should.. I believe the behavior dropping db hasn't been fixed yet.