Wow. Because the data just come “from somewhere”. Where exactly do they think?
I guess that’s a symptom of not being “technical”
Oh really? I know they had their own data of sorts, but you’ll see we have other threads on OSM here. Lots of people have asked us to “do something” to make OCM part of OSM but I never really got past the idea of trying to jam high fidelity data into (slightly more ramshackle) a low fidelity metadata format.
I think we need to get PlugShare onboard before it’s too late.
Well that and they don’t really care as long as they don’t have to look in multiple places. PlugShare have recently started adding ads to the web UI which is symptomatic of seeking a revenue stream, and in turn will likely play out into either a subscription or finding cost savings. They have changed hands 3 times in the past 10 years so eventually they will have commercial pressure, using us for data would be a potential cost saving for them and would centralise edits/imports for the community. I put a similar idea to Chargeway a while back but that fell down due to mutual lack of interest.
That leads me back to the data model, governance, SLAs, quality etc. I doubt in our current state that our data (or service) is attractive to them. I think there is a solution but it’s quite a significant task:
- Attracting high quality data from networks and implementing it in a standard way with least friction for us or them.
- Enabling more data quality workflows (edits, reviews, feedback)
- Formal governance (not really my bag, needs keen and committed communicators).
- Providing a robust service. This may have to involve networks contributing a subscription fee relative to their network size (they get advertising of their network after all). Maybe large apps/services and other consumers should be on a subscription as well. I’m reticent to introduce these, then I remember the ~$20K spent so far hosting OCM
Ok, so data model it is!
Do we:
- Create OCPI classes/tables, and then import/transfer/copy all data in, or
- Create OCPI data feed(s) to make the data externally compliant to OCPI, or
- Mirror all data in OCPI format, then just BANG! turn one off, then other on, or
- Option 2, then option 1, then remove all old tables and classes
I don’t like option 3, but it’s a valid option
My pragmatic idea (only really stubbed and with a dummy Test) was to create an OCPI model and something to convert to/from:
It became clear that we would lose data in the conversion, which led me back to thinking about our data model. I do think we can accommodate most of the important data items with minor changes.