Is it real to get unique(or 80% unique) values for for faker provider in output?

[alex-tomson]

I'm getting the same name for entire column when use faker.first_name. Is it real to get different values to each row?
Maybe I'm doing something wrong, so give me advise please, how I can get different values for faker provider

[alex-tomson]

After some time diving in codebase I find quick solution
In file pganonymize/utils.py change line

data = parmap.map(process_row, records, columns, excludes, pm_pbar=verbose)

to

data = parmap.map(process_row, records, columns, excludes, pm_pbar=verbose, pm_parallel=False)

Seems like force parallelization going bad for faker provider

[hkage]

Hi, sorry I didn't have the time to take a look into your issue. I must admit that most of the recent changes, including the usage of parmap, are features I haven't tested so much, because we still don't use pganonymize in production and I can only develop and improve this project in my spare time.

Afaik pm_parallel=False will deactivate the parallelization, which completely negotiates the usage of parmap. Therefore I cannot simply disable it for now. I think I will have to take a deeper look into faker in combination with parallelization, otherwise I will have to add an option to disable it.

EDIT: Issue #31 describes the same behaviour, so it seems to be a general issue with faker and parmap.