getGPS fails if you convert ts to POSIX before running as in example

[plorch]

If you follow the examples in 5.5.2 and 5.5.3, and run getGPS(), you will get errors because getGPS() expects ts to not be POSIX. You can fix this by running something like the code below first (leaving off the part converting the ts to POSIX timestamps from the code block immediately above the gps_index example).

df.alltags.sub <- tbl.filtered.alltags %>% 
  filter(probability == 1) %>%
  collect() %>%
  as.data.frame()

This highlights an inconsistency that many functions need ts in timestamp format, but not getGPS()

This is also related to issue #36 in that there are other places in the book where df.alltags.sub is generated without converting ts to a POSIX timestamp.

[StuMackenzie]

@steffilazerte

[steffilazerte]

Hi @plorch thanks for your troubleshooting!

We're in the process of revamping and updating the documentation, so a lot of it will change.

This is a good thing to include, and I'll definitely make it clearer in the getGPS() errors and documentation what the requirements are.

Timestamps are sometimes numeric because this allows us to manipulate the data before making it flat which can be much faster for larger data sets.

Thanks!

[plorch]

Thanks Seffi. I understand the why, but most people will stumble at this critical step without a quick fix. I submitted a push request. I got some crazy GPS data once I got it working using the ‘daily’ option: Tags in California, gpsLat and gpsLon on Ohio. Trying the much slower ‘closer’ option overnight. -Pat ***@***.***

…

On Jan 19, 2022, at 3:22 PM, Steffi LaZerte ***@***.***> wrote:  Hi @plorch thanks for your troubleshooting! We're in the process of revamping and updating the documentation, so a lot of it will change. This is a good thing to include, and I'll definitely make it clearer in the getGPS() errors and documentation what the requirements are. Timestamps are sometimes numeric because this allows us to manipulate the data before making it flat which can be much faster for larger data sets. Thanks! — Reply to this email directly, view it on GitHub, or unsubscribe. You are receiving this because you were mentioned.