You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, developing modules is fairly painful. My workflow has been to developing on a fork and making bug fixes by creating releases, which is tedious and fraught with potential errors.
The process could be greatly simplified if we added functionality to support installation to and from disk.
I'm imagining an API similar to remotes::install_github/install.
In this context a user would download a module release by sampling executing commands like:
Strategus::install_github("OHDSI/CohortGenerator") # get latest release
Strategus::install_github("OHDSI/CohortGenerator", ref = "develop") # get development version
Strategus::install_github("OHDSI/CohortGenerator", ref = "v0.1.1") # get specific version
Strategus::install_local("./PLPWithCustomisations") # install from disk
These could then save to disk in a location that is either in the project for re-use or within some global space configurable with an environment variable.
The install process could also optionally instantiate the modules.
Form this we could also add functionality such as:
Strategus::getAvailableModules() # Returns a list of all the currently installed modules with their version numbers
I would also like to see module code somewhat isolated from the R environment so instead of:
# bad
source("https:://pathToGithubUrl/ModuleSettings.R")
createModuleSepecifications(...)
We could implement a function that returns a locally sourced environment:
I also note that this would allow the creation of common interfaces around module settings functions instead of createCohortGeneratorModuleSpecifications it would be a common function createModuleSepecifications.
The text was updated successfully, but these errors were encountered:
Currently, developing modules is fairly painful. My workflow has been to developing on a fork and making bug fixes by creating releases, which is tedious and fraught with potential errors.
The process could be greatly simplified if we added functionality to support installation to and from disk.
I'm imagining an API similar to
remotes::install_github/install
.In this context a user would download a module release by sampling executing commands like:
These could then save to disk in a location that is either in the project for re-use or within some global space configurable with an environment variable.
The install process could also optionally instantiate the modules.
Form this we could also add functionality such as:
I would also like to see module code somewhat isolated from the R environment so instead of:
We could implement a function that returns a locally sourced environment:
I also note that this would allow the creation of common interfaces around module settings functions instead of
createCohortGeneratorModuleSpecifications
it would be a common functioncreateModuleSepecifications
.The text was updated successfully, but these errors were encountered: