the objects in data are currently not available on a system, if the package is devtools::install()ed, and also not during testing.
It appears that the data is only available as exported objects after devtools::load_all().

So that appears to be a reproducibility bug.

Because a docker image, and by extension Azure requires a properly installed and working package, this currently makes it impossible to deploy the app.

add select option to include all institutions in scatterplot?

depends on https://github.com/subugoe/muggle/issues/124

if you want to @Ahobert and @njahn82 I'd be happy to migrate this over to the muggle setup, so we can have this online as a proper vignette as well.

That would be a quick fix for the problem of making this public #9: we can then have the reproducible code online (via echo = TRUE), but not the actual full git source.

I can't promise this would work quickly though, since that requires the project to actually be reproducible.

Also, relatedly, is there by any chance already a package for which this might be a vignette?

would be great to reproduce these.
Made from data-raw/export_tables/

add select option for axes, i.e. possibility to plot total oa share vs. fully oa journal share or something like that

recreate data gathering from WoS data bases

the article version of the preprint currently only has hard-copied citations; they're not run through pandoc citeproc and are not included in the *.bib.

includes table 1 from preprint

In the preprint, this is written in Python.
@Ahobert can you just paste the code from your colleague right here in the comment?

I think it might be nice to actually have these in one repo, and one R package, with the paper as a vignette, and the dashboard as another vignette.

this can be reproduced from open sources, though they may not be available all the time, so we should git lfs this as well.

came up in subugoe/oauni_10_18#24

depends on #8

@Ahobert will share the file with me.
This file is the source for the different (publishable) aggregations.

pubs_cat.csv is created via the sql calls from proprietary datasets, which requires (personalised!) access credentials only available for some team members.
The pubs_cat.csv itself still includes proprietary data and must not be published.

same as #8

BFG: https://rtyley.github.io/bfg-repo-cleaner/

TLDR for all the poor souls I had to tag now: @Ahobert asked me to purge some data from this repo, before it could go public.
This is done, but y'all will have to delete and re-clone it (see below comment).

• minutes.md

articles is number of articles per institute, not per sector, which we used to order by in the paper. Does not seem to change the ordering for our dataset, though. https://github.com/subugoe/oauni_10_18/blob/c5a331954d410428e05bba9d81aa41272d721b85/data-raw/select_inst_sector.R#L6

Error in value[[3L]](cond) : 
  No root directory found in /srv/connect/apps/oadash-refs-heads-muggleme or its parent directories. Root criterion: contains a file `DESCRIPTION`
Calls: local ... tryCatch -> tryCatchList -> tryCatchOne -> <Anonymous>
Execution halted

Might be related to golem ThinkR-open/golem#387 (fix may not be on CRAN)

while doing that we'll also remove the pk_kb_inst column, because that is actually somewhat proprietary (?) but in any event unnecessary info, because the downstream antijoins only use the second column.
So also covers which can be closed #28.

So it's just a string.

the rda can go, but some rows should be kept as snapshot tests

from what I can see now, this could probably all be done with crosstalk, obviating the need for a shiny runtime.
And it might be worth to standardise on one dashboard solution.
I think flexdashboard is, at this point, the forerunner, because it is more actively maintained and is more lightweight.

once they are chunks

http://styler.r-lib.org

Don't want to do this now, because that would cause big diffs and merge conflicts and confuse everyone.

this code still has a lot of notes and warnings, though they're not urgent.