Closed salvafern closed 1 year ago
This is my personal opinion, not necessary the opinion of the whole editorial board :wink:
To me if you have a clear separation of functionality and dependencies between the potential two packages, a split sounds good. Reasons I think a split sounds good: easier for you to maintain, easier for an external contributor (or a reviewer) to get acquainted with the codebase.
With maybe an umbrella package (like taxize / tidyverse / devtools) that'd call the other two packages?
Thanks a lot @maelle ! We will most likely do the split then and add an umbrella package as you suggested. I hope we can submit to rOpenSci within the next weeks.
Discussed internally with part of the MR Team (@LennertSchepers & @brittlnv). We won't split the package for now. First we will make sure if there are any plans on ows4R
to upgrade to httr2
and if not we may write our own WFS calls and retrieve via httr2
.
This would have even some advantages as e.g. we can download to disk directly and use a output format more efficient than GML.
We have decided internally to leave it as is for now as the tests are working (see #16). We will revisit this during the rOpenSci review, maybe there will be new ideas raised.
As mentioned before, there are two well distinguished parts in this package with different functions:
gaz_search("Belgian Part of the North Sea)
mrp_get("eez_boundaries")
Some details:
httr2
for HTTP calls.rdflib
to deal with RDF datahttptest2
to mock up HTTP responsesows4R
, which depends inhttr
and not inhttr2
leaflet
+leaflet.extras2
for WMS visualization (moved to suggests and added check to avoid package failure in case of missing library leaflet)httptest
to mock up HTTP responsesThis brings a problem: we used the latest
httr2
for the gazetteer, but due toows4R
we also need to depend onhttr
And because we depend both on
httr
andhttr2
, we need to use bothhttptest
andhttptest2
to mock up calls. This brought me some headaches as the two packages tend to overlap, but it is solved by cleaning after the use of each test package. E.g. detach and set requesters / or redactors to NULL in each test. See https://github.com/lifewatch/mregions2/commit/fa2acce90e314821d391fe90dc3dbdcd14a4f6d2I see three options that make sense here. Either:
httr
instead ofhttr2
, orows4R
As we are planning to submit this to ROpenSci, @maelle Could I ask you to quickly give your opinion on what's the best way forward in this case, so I can do the necessary changes before submitting? Thanks a lot in advance 😊