Towards the end of August, I received a "Proficiency Evaluation", which is the final step in getting production keys for the Oregon Metrc API. Because this is essentially a test, I'm not going to share any of the questions (or answers). However, this was my first chance to play with the API, so I will share my impressions.
It took me over a month to complete the evaluation. And it wasn't because I didn't invest the time! The initial set of docs I got were . . . well, wrong. A number of the argument names weren't correct (though a few were easily guessable). A few of the endpoints just flat-out didn't exist. Of course, none of these issues are immediately obvious, so each of them wasted a lot of time.
Feeling defeated, I contacted Metrc to explain the problems I was having in completing the test. I can not over emphasize how helpful and responsive Metrc has been (Michelle is my hero and I could not have gotten to this point without her help!). I was pointed to some updated online docs. Much better!!!
However, it still wasn't smooth sailing from there.
To complete the evaluation, I was granted access to a sandbox. I was sent a list of plant and package tags, for a number of different licensees, and the sandbox has some initial data. It eventually occurred to me, not all (or maybe not any) of the data was created by the good folks at Metrc. The sandbox, the credentials, the list of tags . . . everything was shared with everyone else who was trying to complete the evaluation. This makes for some nasty surprises and confusing moments!
The shared tags made it impossible to know which tags had been used and which could be attached to plants or packages. To find usable tags, I ended up looking them up until I got a non-200 response. As time went on, finding unused tags became harder and harder. I'm pretty sure I ended up using the last plant tag for one of the producer accounts.
It would be nice to believe you can create something, like a plantbatch, and walk it through it's lifecycle. But in a shared environment, especially one in which users have no means of communicating with each other, your data can get stomped at any time. I had data change out from under me. Initially I felt a tinge of guilt when I used other peoples objects. But by the end, I was flagrantly harvesting other peoples plants and selling out of other peoples packages.
Overall, the API makes a lot of sense. As long as you understand the Metrc system, figuring out which endpoints to call is pretty intuitive. I like the RESTful approach, and I think this is a pretty good implementation.
There are some rough edges though. There's some inconsistency in some of the endpoints. For example, in one case you "changegrowthphase" and in the other case you "changegrowthphases". You can "destroy" a plantbatch, but individual plants require "destroyplants". And while you tag new plants using a "StartingTag", you find plants with a "Label" (these are the same thing). The part I like least is the difference between shapes of requests and responses. For example, when you create a plantbatch, you give it the "Strain". But getting a plantbatch returns the "StrainId" and "StrainName".
I found there's very limited access to historical data. So, if you change the growth phase on a plantbatch, and later see the strain is different, or it's been harvested (or both in my case), there's really no way (that I could tell) to figure out what happened to the plants. Of course, if you're in an environment with a single licensee, none of this should happen . . . but if you wanted to show the lifecycle of a plant, you're going to have to jump through some hoops.
I was less than thrilled to find successful POST requests always return empty responses. Non-200 responses include pretty good information, but after making a successful call, you just get a 200. This pretty much doubles the number of calls I ended making . . . create something, then go look it up.
Though there were some frustrating moments, all in all this was a pretty fun challenge. By the end of completing the evaluation, I felt pretty comfortable with the API and look forward to actually building some software that leverages Metrc. I'm secretly hoping some details of the API are still a work in progress, but even if it's fully baked, I can envision building a lot of cool integrations.
