pierucci/rgho

:globe_with_meridians:Access WHO Global Health Observatory Data From R

rgho's Issues

Cannot install rgho on R 3.4

Error : object ‘timeout’ is not exported by 'namespace:memoise'
ERROR: lazy loading failed for package ‘rgho’

Do not memoise get_gho_data

See #16.

Do not do that before pierucci/heemod#119 is on CRAN.

Combine help files

Write tests

Needed to start #7 (check it doesn't introduces regressions).

Hide documentation of unexported functions

With @keywords internal.

Control get_gho arguments

Maybe using option().

Create rgho.R file

Move imports.R there and add package description.

Refactor tests to be insensitive to GHO changes

Test fails when codes are added

Should test that length > something.

Loop in get_gho doesnt loop.

Always changes proxy settings after the first trial.

Use codecov

Follow httr API best practices

https://cran.r-project.org/web/packages/httr/vignettes/api-packages.html

Rename functions in a more consistent manner

Add methods for gho objects

http://adv-r.had.co.nz/S3.html

When implementing a vector class, you should implement these methods: length, [, [<-, [[, [[<-, c. (If [ is implemented rev, head, and tail should all work).

More informative error messages

So that errors in packages relying on rgho can be more easily tracked (see pierucci/heemod#118).

Document options and memoisation

Use R default proxy settings

Like in install.packages().

get_gho_codes fails for some dimensions

get_gho_codes("DATASOURCE")

Error: length(x) == length(labels) is not TRUE

Migrate from XML to json

Because:

1. XML is a pain.
2. xml2 keeps crashing in strange ways.

Fix: add format=json to API calls.

Use modify_url in URL building step

Instead of the mess right now...

Memoise for a limited time

The GHO data can change with time, so memoised functions should forget their result after a while to avoid returning stale data.

This is only an issue on sessions that last for days.

memoise(a, ~timeout(10))

Benchmark if get_who_data() memoisation is necessary

It shouldn't change much because requests are already memoised. But on the other hand fast execution is necessary for heemod.

The slow part may be in the XML parsing.

Print N option

Report occurrences of faulty json

In clean_json(), when verbose = TRUE.

Sort vignettes

With prefix.

Dont memoise errors

Don't memoise result if request returns an error.