search

openfarmcc / Crops

Discussion on how to best manage a Crops data provider so that it's most widely useable.
Other
22 stars 1 forks source link

Live discussion Feb 10, 2016 @ 6pm UTC #11

Open mstenta opened 8 years ago

mstenta commented 8 years ago

Hi everyone! We've had some great conversation in these issue queues around the crops data schema. I was wondering if any of you would be interested in doing a Google Hangout to discuss it some more face-to-face?

I host a monthly farmOS development call via Google Hangouts, and I thought this would be a great topic for the next one.

I haven't set a schedule for the 2016 calls yet, but last year they were the first Wednesday of every month at 7pm EST. However, since many of you are in Europe perhaps we could shift that to better accommodate?

What do you think about Wednesday, February 3rd at 1pm EST, which is 6pm GMT (I think?). Or if that's too soon - Feb 10th?

mstenta commented 8 years ago

@simonv3 @andru @roryaronson @pmackay @Mageistral @tkeifer - would you be interested in joining? (sorry if I missed anyone - feel free to tag more!)

Mageistral commented 8 years ago

As there is some interested people, I was thinking the same, at least to see who we are :) or maybe you already know each other. I'm completely unavailable the 3rd so the 10th is better. And if understand properly the time, that's ok, 7pm /France.

roryaronson commented 8 years ago

I can do the 10th!

mstenta commented 8 years ago

Great! I've also heard back from a few other interested folks and it sounds like they can do the 10th at 1pm EST / 6pm UTC as well.

I posted the 2016 farmOS Monthly Call schedule here, where you can also find the link: http://farmos.org/monthly-call

Looking forward to talking! :-)

mstenta commented 8 years ago

I'll close this ticket so it doesn't crowd the queue...

simonv3 commented 8 years ago

How about we leave it open until it happens - that way people who might be interested will find it easier. How about changing the title to when and where?

mstenta commented 8 years ago

Sounds good!

mstenta commented 8 years ago

Just a reminder to anyone tagged in this thread: the call is scheduled for tomorrow at 1pm EST / 6pm UTC.

The link to join the Google Hangout is available here: http://farmos.org/monthly-call

Hope to see you there! :-)

mstenta commented 8 years ago

I started an agenda for the call in a Hackpad: https://farmos.hackpad.com/farmOS-Monthly-Call-ExlBTwu96ZX

I figure we can start with some introductions, then jump into the crops discussion with a review of some of the Github conversations. Then take it from there and hopefully figure out next steps.

Feel free to add anything in particular you'd like to talk about!

mstenta commented 8 years ago

@roryaronson @Mageistral Thanks for joining the call! Here's a quick summary for those who couldn't make it:

We started with introductions, and went through some of the user stories described in #9. Then we reviewed some of the discussions around schema definition, data format, centralized vs distributed considerations, separations of concern, etc.

We had some folks on the call who weren't familiar with the discussions, so it was helpful to bounce ideas off of fresh ears, so to speak. And just great to meet and talk face to face in general.

We didn't come to any firm conclusions or decisions on anything, nor was I hoping/expecting to. The general agreement was that we all need this, and there's good use-cases for others as well. And that maybe the best approach moving forward is to just start building things in our own respective domains and applications, and keep these common schema and format conversations going at the same time - so we can share ideas, experiences, and requirements as it develops.

In some ways, we are in a stage of "which came first: the chicken or the egg" - we need a common data definition to build applications on top of, and we need applications to define the common data definitions. It makes sense that they would evolve together. :-)

I was hoping to keep more detailed minutes during the call, but alas I failed at doing that. This was just from memory, so @Mageistral and @roryaronson if there's anything I missed feel free to chime in!

andru commented 8 years ago

Sorry I missed it folks. I've been heads down and lost track of it.

I'm prototyping rapidly with Hortomatic so my short term approach is to pull together a dataset that works for my immediate needs. In that sense I'm on the same page as the general consensus of the call. When I've got something going I'll publish it and share it here to feed into the discussion.

roryaronson commented 8 years ago

I think that sums it up @mstenta. And thanks for organizing the call - it was nice to have that real-time discussion and see some new faces :)

One other thing we talked about was starting very small, and very simple. Using git would be the easiest way to start and open up the door to other developers. Using something like a Dat database or other technology poses a potential barrier and technological complexity that we don't need right off the bat.

I think everyone agrees that at the bare minimum this shared database requires crop names and unique identifiers. We could talk for long lengths about what else to add beyond that, but as pointed out above, that's sort of a chicken and the egg thing. It seems best to move forward with something as simple as just names (which we already have a lot of data for too) and see what happens.

As all of our projects and this shared resource develop, the path to move forward will likely become clearer as far as how and where to host/access the data, what else to add to it, how forking and merging should work, and how a developer can (eventually) use data from multiple sources that are sharing the schema. (eg: growing info from OpenFarm, medicinal usage data from Practical Plants)

Mageistral commented 8 years ago

Hello, a quick word to go on the same idea. Maybe tools like Dat data are great but for sure, it will have side effects. For example, I'm not sure it's easy to use it with every hosting service provider.

In the data flow, it's obvious we will have data flowing from our app to the DB, isn't it ? This way, we must choose pivot format first (?)

mstenta commented 8 years ago

@Mageistral can you elaborate a little? I'm not sure what you mean by pivot format, exactly.

I think the goal is to define a data format that is 100% independent of any specific application or database. This will allow data to be shared between very different applications, using different technologies, databases, etc.

A secondary goal might be to build tools that help integrate this data format into specific use-cases - but that should be a separate goal from the definition of the data formats themselves. So something like Dat Data might be helpful for someone to use, but it should not be part of the core data definition, in my opinion. The core data definition and schema should be agnostic to the technologies that use it.

Does that make sense? Apologies if I misunderstood your point.

Mageistral commented 8 years ago

During the talks and definitions, yes, structure and data should be agnostic to the technologies. The thing is we almost agreed on a structure object crop with a label and uuid attributes Writing my comment, I forgot this https://github.com/farmOS/CropDB-Spec So, my points are already answered... Now, we need to do this very small spec + YAML sample ; if I have some time this w-e I'll. And think about how to update this data, from outer source ? From inner app ? How to do deal with it without Master Data Management huge tool.

mstenta commented 8 years ago

And think about how to update this data, from outer source ? From inner app ? How to do deal with it without Master Data Management huge tool.

This is a good question. My opinion is that the YAML files would be considered "official" sources of data - and they would only be editable via Git. But... within any application that uses the data, it could be handled in many ways, depending on what the application needs.