Possible issues with bringing sapfluxnetQC1 up to date

sapfluxnet / sapfluxnetQC1

R package with functions related to sapfluxnet project

6 stars 0 forks source link

Possible issues with bringing sapfluxnetQC1 up to date #47

Open MalditoBarbudo opened 2 months ago

MalditoBarbudo commented 2 months ago

[x] Spatial packages: sapfluxnetQC1 is based in sp, rgdal and rgeos when using spatial objects (coordinate checks...). This will likely fail in a recent R installation. rgeos is also retired.
[x] GSODRdata package used to get wordclim values at the nearest stations has been retired. We need a new way to obtain these values to calculate the site biome.
[x] tpl package for fixing species names is also unmaintaned, we need to implement an alternative
[x] methods for SfnData class must be fixed, as some of them fail (i.e. [). Or, just implement the class in sapfluxnetr here, if trivial.
[x] Fix failing tests
- [x] deprecation notices
- [x] others

rafapoyatos commented 2 months ago

For the taxonomy, check the packages I mention in issue #46

MalditoBarbudo commented 2 months ago

Deprecated packages (sp, rgeos...) fixed. Methods for SfnData class fixed.

A lot of tests failing, due to different reasons (tests based in removed packages and output, tests issuing unexpected warnings due to deprecation notices, tests failing for changed behaviours in dependencies). Working on it.

MalditoBarbudo commented 2 months ago

All tests passing :)

vflo commented 2 months ago

All the QC1 and QC2 works fine with ESP_VAL_SOR test dataset.

vflo commented 2 months ago

I've included a check to determine if the coordinates are on land. However, @MalditoBarbudo, I'm not sure how to include a test process.

vflo commented 2 months ago

test: 'wrong coordinates are plotted correctly' doesn't work properly. I'm not saving the coordinates plots now, just printed in the report.

MalditoBarbudo commented 1 month ago

I've included a check to determine if the coordinates are on land. However, @MalditoBarbudo, I'm not sure how to include a test process.

I've been looking at the changes, and after all the process, new messages are shown when the plot is located in water. In the tests for coordinates, in B2 is where I tested the is inside country trait. Basically I created bad coordinates and good coordinates and expect the correct messages. In this case I would add a set of coordinates that simulate the on_water trait and check for the presence of the correct message, as well as the correct number of bad coordinate sites.

MalditoBarbudo commented 1 month ago

test: 'wrong coordinates are plotted correctly' doesn't work properly. I'm not saving the coordinates plots now, just printed in the report.

Not sure what is happening here, but changes were made to the plot code as well as forcing the plot argument. Could be worth to check the code there.

rafapoyatos commented 3 weeks ago

I successfully used the 'devel' QC1 with GBRDEV* datasets! @vflo Just one comment on the species checks @MalditoBarbudo . What happens when a contributor only defines taxa at genus or even family levels? Can we detect this and use some common terminology: e.g. 'Quercus sp.' when somebody reports only 'Quercus' or 'Quercus spp.' (incorrectly). What about family-level? Could we use e.g. 'Myrtaceae fam.'? See here , for example.

MalditoBarbudo commented 3 weeks ago

@rafapoyatos Not sure what happense in that case. I'll need to revisit the species checks to see if is possible or already done by one of the dependencies used.