Closed ungoldman closed 9 years ago
@chelm this explains a lot re: that weirdness we were dealing with a few weeks ago. Using 1.0.0+ (and starting new modules at 1.0.0) should allow us to avoid dealing with versioning issues like that.
I updated providers.md, caches.md, & plugins.md to make it a lot easier to see version and build status for every module related to the project.
Here's everything under the koopjs org.
version | module |
---|---|
ArcGIS Online | |
GitHub | |
GitHub Gist | |
Socrata | |
CKAN | |
American Community Survey | |
OpenStreetMap | |
GeoCommons | |
VRBO | |
Census | |
Yelp | |
PostGIS | |
ElasticSearch | |
Koop Tile Plugin |
Shouldn't we only do this for packages we consider stable?
E.g. Yelp is totally expiremental and ES cache is not finished.
@dmfenton Depends on your release philosophy. I would argue we should start at 1.0.0 so our version numbers actually mean something with respect to SemVer and npm
. See https://nodesource.com/blog/semver-tilde-and-caret#recommendation-start-at-1-0-0, namely advice from two major NPM developers:
We can (and should) mark projects that are unstable as alpha or beta (or unfinished, or experimental) in the documentation (README). If we use 1.0.0+, when we fix bugs and upgrade the modules in different ways, it will get pulled in by projects that depend on them. :smile_cat:
Focusing exclusively on officially supported modules, which are now listed in the readme (https://github.com/koopjs/koopjs.github.io/#providers).
Everything's 1.0.0 or above! yay :tada: :balloon:
See https://nodesource.com/blog/semver-tilde-and-caret, especially https://nodesource.com/blog/semver-tilde-and-caret#1-0-0-anxiety. Using 1.0.0+ versions simplifies dependency management and ensures we are really using SemVer.
^0.y.z
will always lock to that specific number, wheras^1.y.z
will pick the highest minor and patch version without going above 1, which I think is a lot easier to deal with and what people expect.