cont-limno / lagosne Goto Github PK

@jsta -- I just tried to get Emily Stanley (@ehstanley) set up with the R package and she doesn't have access. How do I change this?

Tables "lakes4ha.buffer500m" and "lakes4ha.buffer100m" do not import with column names (column names in row 1).

This was present in the legacy code base.

See http://r-pkgs.had.co.nz/vignettes.html

Refactoring the code will need to be done carefully.

The vignettes as currently written require that the full LAGOS dataset is installed and available to lagos_load. I think this is good because it allows us to describe the contents of the data product. However, automated build testing via CI services (and eventually CRAN) breaks because the they don't (and probably should not) have the data available. The only solution I can come up with is to make the vignettes static. That is to pre-build all the vignette figures and tables and display the code chunks without running them (eval = FALSE).

Need to sanitize inputs

In tables lakes4ha_buffer100m and lakes4ha_buffer500m, column is "lagoslakeid" after importing through package, but column name in geo v1.040 is "lakes4ha_buffer100m_lagoslakeid" and "lakes4ha_buffer100m_lagoslakeid".

See #15

Some functions have a title in the documentation and other do not. Consider following the rules at: http://style.tidyverse.org/code-documentation.html

Something like the first 10 lines of every table

the data compilation functions will need to be updated

The keywords in LAGOS:::keyword_partial_key() are not included in any of the user-facing documentation.

For example, ?locus does not yield any results.

Also buffer100m.lulc buffer500m.lulc lakes.geo lagos_source_program

At the very least, make preprocessing functions operate on column names and not column numbers

A lot of this code may have reinvented the wheel. It is likely that we can simplify a lot of this by wrapping the dplyr package.

LAGOS:::lagos_compile(version = "1.054.1", format = "rds")

fails with error message:
Error in gzfile(file, mode) : cannot open the connection
In addition: Warning message:
In gzfile(file, mode) :
cannot open compressed file 'C:\Users\Samantha\AppData\Local\LAGOS\LAGOS/data_1.054.1.rds', probable reason 'No such file or directory'

R version 3.3.1 (2016-06-21)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 7 x64 (build 7601) Service Pack 1

sessionInfo()
locale:
[1] LC_COLLATE=English_United States.1252 LC_CTYPE=English_United States.1252
[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C
[5] LC_TIME=English_United States.1252

attached base packages:
[1] stats graphics grDevices utils datasets methods base

other attached packages:
[1] LAGOS_1.054.1

loaded via a namespace (and not attached):
[1] lazyeval_0.2.0 magrittr_1.5 R6_2.1.2 assertthat_0.1 rsconnect_0.4.3
[6] DBI_0.5-12 tools_3.3.1 dplyr_0.5.0 rappdirs_0.3.1 tibble_1.1
[11] Rcpp_0.12.7

rappdirs::user_data_dir("LAGOS")

"C:\Users\Samantha\AppData\Local\LAGOS\LAGOS"

In addition to the ability select columns by name, allow the user to select pre-defined groups of columns within each table. For example, "atmospheric deposition" that may include multiple variables.

This is the bulk of the old LAGOS R package but it could be much cleaner by using functions to reduce redundancy versus copy-paste.

Store data using https://github.com/hadley/rappdirs

Possibly implement some of the ideas in https://github.com/richfitz/datastorr/blob/master/vignettes/datastorr.Rmd

epi.nutr -> epi_nutr
lakes.limno -> lakes_limno

exactly which ones is an open question

https://portal.lternet.edu/nis/mapbrowse?packageid=knb-lter-ntl.320.4

http://datadryad.org/resource/doi:10.5061/dryad.75s9s

sampling event, lagoslakeid, etc.

LAke multi-scaled GeOSpatial & temporal database -> Tools for Interacting with the Lake Multi-scaled Geospatial and Temporal Database

• @jiayuzhou can you do this? I cannot because I do not have administrator level control of the repository.

@jsta where is the best place for this information? As a user, I could imagine wanting this to be in two places: 1) some sort of documentation listing each variable within each table, along with some metadata (units, plain English description, etc). We could have documentation for each table, but where (in the package structure) would this go? 2) in a table format, similar to the info table that is currently imported with the rds file.

Add checks to see if data are already loaded. This is what datastorr/storr does well.

Should this be done at the table level or the module level?

A nice feature of the package might be to compile ALL previously published LAGOS data (I think all of it is on the LTER portal). So the function could be something like lagos_published(paper = c("Oliver2015", "Lottig 2014"))

See http://r-pkgs.had.co.nz/inst.html#inst-citation

We might consider adding additional alias terms relative to the datasets. For example, we might list "chla", "colora", and "doc" as aliases for the epi.nutr table. This would enable ??LAGOS::chla

See #22, and the output of devtools::test(). Need to throw helpful error messages when queries try to return data that does not exist.

This was done to be consistent with the legacy loading scripts but can be a real pain for analysis. Is there any reason not to load as character strings instead?

One idea might be to use the units package

Inquiring minds want to know.