deduplicate() and extract_domain()

[sebstier]

Hi,

the new version of deduplicate() returns a data.frame instead of a wt_dt object.

the new version of extract_domain() with the adaR implementation does not accept drop_na = FALSE anymore, but keeps all rows by default. That is intentional, right?

[schochastics]

deduplicate is fixed now.

The second was intentional, but probably should have a proper note. The reasons to drop the parameter are twofold:

1. adaR fails less (actually almost never) to parse the domain and
2. if we want to drop certain rows then we should write a proper filter function

[sebstier]

Thanks. Two more points on extract_domain():

• Examples are still tailored towards the previous version
• extract_domain() only seems to extract a valid domain when a url starts with http/https. That is sometimes not the case, e.g. in the 6-country 2019 data.

[schochastics]

The open pull request #95 is going to deal with those issues. Just had no time to finalize it yet

[sebstier]

two more issues:

• deduplicate() creates two timestamp columns
• deduplicate() converts format of column timestamp from dttm to dbl

[schochastics]

thanks, thats a bug. I will fix that in the open PR

[schochastics]

@sebstier can you post a snippet that produces this error? cannot reproduce it
nvm, found an example. Happens for method=aggregation