Is it guaranteed that if dvc status Dvcfile shows: nothing to reproduce, the dvc repro Dvcfile will not do anything ?
For my use case, it looks like dvc repro (even when there is nothing to reproduce) takes more time to run than dvc status. I could then switch to calling dvc status to find out if there’s nothing to reproduce and have a faster pipeline.
Hi @tmain !
Sorry for a late response.
Is it guaranteed that if
dvc status Dvcfileshows: nothing to reproduce, thedvc repro Dvcfilewill not do anything ?
Yes, it is indeed guaranteed.
For my use case, it looks like
dvc repro(even when there is nothing to reproduce) takes more time to run thandvc status. I could then switch to callingdvc statusto find out if there’s nothing to reproduce and have a faster pipeline.
Repro works in a bit different way than the status does so there might be differences in processing time. Sorry, I’m not sure I follow, could you please elaborate on what you mean by switch?
Thanks,
Ruslan
By switch I meant the following.
In order to find out whether I need to repro something, I used to call dvc repro Dvcfile --dry and parse the output of that command. If I got Nothing to reproduce, I would continue without calling dvc repro Dvcfile.
Now instead, I can do dvc status Dvcfile, find out whether there is something to reproduce, and call dvc repro Dvcfile if there is.
Ah, got it. Thank you for clarifying! Indeed, dvc status was intended to be used precisely for checking if anything in your pipeline has changed and requires reproduction. And repro --dry is actually intended for visualizing the order and list of the commands that will be run in order to reproduce the pipeline 