search

ropensci / RNeXML

Implementing semantically rich NeXML I/O in R
https://docs.ropensci.org/RNeXML
Other
13 stars 9 forks source link

Reconfigure (or disable) manuscript rebuilds on Circle CI #211

Closed hlapp closed 5 years ago

hlapp commented 5 years ago

Maybe it's worth reconfiguring how the manuscript rebuild on Circle CI gets triggered. Maybe this shouldn't need to be done for each pull request and each update to it, but only when master changes? Given the shape of the test suite plus the vignettes, it's hardly imaginable what kind of problem the manuscript code would detect that the tests and vignettes missed.

Originally posted by @hlapp in https://github.com/ropensci/RNeXML/pull/210/comment#issuecomment-441111793

hlapp commented 5 years ago

BTW it might well be worth considering to move the manuscript into a separate repo and tie it to a specific version of RNeXML anyway (which I think we cannot do while keeping it in the same repo?). Even if this error looks entirely spurious to me, the time will come that we want to make a change that's not compatible with the code as written in the manuscript, and I'd have reservations about then updating the manuscript unless we'd have some kind of very obvious way to version it. By "very obvious" I mean that people can hardly miss the fact that the version is not the same as the one used to produce the originally published manuscript.

Originally posted by @hlapp in https://github.com/ropensci/RNeXML/pull/199#issuecomment-441085786

hlapp commented 5 years ago

We could move the manuscript to a separate repo, though there might confuse some who expect that it is here. Since it is built on a separate CI system I'm not sure what is gained by doing that. We can modify the Circle-CI settings to test a specific version of RNeXML rather than the current version. It is already using a locked version of R (3.3.1), see circle config. If we replaced the devtools::install() with install.packages("RNeXML", dep=TRUE), it would install the version of RNeXML (and dependencies) that was current when R 3.3.1 was last current. Of course there shouldn't be much need to have that on CI at all then since it would be running the same code base each time, but it would at least be proof-of-principle on the reproducibility; at least within the specified dockerized setup.

Originally posted by @cboettig in https://github.com/ropensci/RNeXML/pull/199#issuecomment-441159023