This Pull Request is obsolete. Please rebase your work on #196 which includes all the items "in the pipeline" below plus numerous other small fixes.
Release candidate for VisionEval 3.0 for R 4.2.2.
What's New
This version manages models, scenarios, results and exports through a simple end-user API. https://docs.visioneval.orgcomes close to documenting this version. To see what "works", run the Github version:
Clone the Github
Start VisionEval-dev.Rproj in the root directory
Run ve.build() to construct the packages
Run ve.run() to launch the runtime (note that the built "runtime" is only used indirectly
VisionEval runs in the new "runtime.test" directory
You can set a directory of your choice selected either by passing it as a parameter (ve.run('myRuntimeDirectory")) or by setting the VE_RUNTIME system environment variable. A complete working runtime will be created at that destination if it does not already exist
Once running, do the walkthrough() function to get a list of sample scripts illustrating basic functions (all to run in an additional temporary runtime to avoid confusing them with real work).
walkthrough() will also work for ordinary users
A more "advanced" set of walkthroughs exists in sources/framework/VEModel but those are yet to be updated to final code changes so they will change before release.
Use the VEModel test functions for a more complete rendition of the API
Run ve.test("VEModel") to load the API test functions (a comprehensive exercise of what works and how).
Inspect the scripts in the walkthrough and the VEModel tests to see how to do various useful things
Chief Accomplishments
Among the many changes made here since the last beta:
works correctly with R 4.2.2 (our main recent hang-up on new releases was an important change to R language behavior)
the query functions are working all the way
all the module packages have been fixed so they pass R CMD check with no messages apart from being grossly oversized
the framework now supports row-level input file consistency checking (not just columns)
many improvements have been made to the run scripts (both for developers and end-users)
previous runs are handled better (including checking for "out of date" status - if their run parameters have changed since the last run)
a great many issues open on VisionEval will be checked and closed as this pull-request is completed (the ones anticipated to be closed are already identified with comments from @jrawbits. I may accidentally have fixed others...
Items in the Pipeline but not here
Things that are NOT in this release (but will be part of the next development branch)
Dynamic module specifications (define Inp, Get and Set with a function call rather than a compiled-in list)
Export function enhancements
generic tabular output, notably to SQL
tabular output of queries that better support Tableau (with revised row/column structure)
Next Steps
I've assigned a couple of helpers to download and test this pull request: please set it up, build it and run it (both using the walkthrough, but also with models of your own). I have successfully run the entire H-GAC 5000 zone model with it, so it probably works. Specific steps:
Once the preliminary testing is complete, this pull request will be merged into development-next, tagged and bundled as a Release Candidate.
We'll ask friends and VisionEval users to try it out then.
Once final fixes have been made, it will all be pushed over to "development" and then to "master" (and perhaps we'll change "master" to "trunk").
Finally, we will retire "development-next" and just go back to having one "development" branch.
jrawbits
commented
1 year ago
This Pull Request is obsolete. Please rebase your work on #196 which includes all the items "in the pipeline" below plus numerous other small fixes.
Release candidate for VisionEval 3.0 for R 4.2.2.
What's New
This version manages models, scenarios, results and exports through a simple end-user API. https://docs.visioneval.orgcomes close to documenting this version. To see what "works", run the Github version:
ve.run('myRuntimeDirectory")) or by setting the VE_RUNTIME system environment variable. A complete working runtime will be created at that destination if it does not already existwalkthrough()function to get a list of sample scripts illustrating basic functions (all to run in an additional temporary runtime to avoid confusing them with real work).walkthrough()will also work for ordinary userssources/framework/VEModelbut those are yet to be updated to final code changes so they will change before release.ve.test("VEModel")to load the API test functions (a comprehensive exercise of what works and how).Chief Accomplishments
Among the many changes made here since the last beta:
Items in the Pipeline but not here
Things that are NOT in this release (but will be part of the next development branch)
Next Steps
I've assigned a couple of helpers to download and test this pull request: please set it up, build it and run it (both using the walkthrough, but also with models of your own). I have successfully run the entire H-GAC 5000 zone model with it, so it probably works. Specific steps: