text stringlengths 0 1.28k |
|---|
We wanna create a platform that allows for plenty of those opportunities for people to create value. We don't want to take business away from people. We actually would rather there be -- we would rather be the platform on which an ecosystem is built, than try and earn money ourselves-- |
**Jerod Santo:** \[24:15\] Take all their ideas and... |
**Coby Chapple:** Right, our endgame is not to be the project management solution; our aim is to create a platform where that discussion around project management, there can be lots of different options that you can use, to pick from when you are starting a new project. Maybe within one company you're gonna use three d... |
**Jerod Santo:** So does Projects have full API support? Can you get at everything that you need to in order to build on top of the Projects feature? |
**Coby Chapple:** At the moment there are... So this is another thing that we announced at the Universe, and that is we are kicking off an Alpha for GraphQL API. This is a big change that we're making in general in terms of our ability to support API access to the product changes we make out of the box when we announce... |
This is something that we've started to actually do with Projects. I would need to clarify online if it's available right now, but certainly within the GraphQL alpha -- it's still pretty alpha... And it's gonna change a lot, too. There's a couple of product features that we've built from the outset using GraphQL intern... |
**Jerod Santo:** So GraphQL is the way of the future... |
**Coby Chapple:** We spent a lot of attention on our API. We understand how important it is to have an API be usable and enjoyable to use, and... |
**Jerod Santo:** Approachable, and discoverable... \[laughs\] |
**Coby Chapple:** Right, exactly. So we've looked at a lot of the things that people have built, like hypermedia... There's a lot of principles that we've explored a lot in terms of how to best do this, and when we stumbled across the GraphQL stuff it was very interesting to us. There's a lot of people at GitHub that s... |
**Jerod Santo:** That's cool. |
**Coby Chapple:** There's a lot of stuff that comes for free with GraphQL that is very interesting to us. We've spoken a lot with the people at Facebook, we're worked very closely with them about this, and we've seen them use it in production for a very long time. I think they've been using it in production since 2012,... |
**Jerod Santo:** Wow. |
**Coby Chapple:** They've been using it internally there with a huge amount of success, and we think this is technology that a lot of other people should look at as well. It's not gonna be something that suits everybody... |
**Jerod Santo:** \[27:58\] Yeah, there are a lot of detractors, a lot of naysayers. I think GitHub is probably the second major adopter of GraphQL, outside of Facebook -- the first one outside of Facebook. So there's people that are super excited... You can say this with anything, right? There's people who are super ex... |
**Coby Chapple:** Right. For me, I look at it as APIs need to be dependable. So this is a decision that I think will actually reduce the amount of risk involved, because it means that if you build something on the GraphQL API, unless you change what your querying, the functionality will continue to work. |
**Jerod Santo:** It's not like an endpoint's gonna disappear on you. |
**Coby Chapple:** Right. So there's a lot of things about GraphQL in terms of versioning clients that actually reduces a bunch of complexity. You no longer have to think about a versioned API as you sure initially might. It actually takes a lot of those concerns away and allows you to build and iterate on clients in a ... |
I think now that we are kind of putting our name out there and saying, "Okay, we're gonna double down on this too", I think that's gonna make a lot of other people prick their ears up and say, "Okay, maybe I should take a more serious look into this", and I think that's gonna start to see more people investing in this ... |
**Jerod Santo:** It seems like a boon for the client-side, but harder on the server side, just from my very - what's the word for newb? - newb understanding of GraphQL. It seems like implementation, from you guys' perspective and from anybody else's as potential adopters of API provider, it seems like a lot goes into t... |
**Coby Chapple:** I think one of the benefits in that side of things though with GraphQL is it doesn't actually dictate anything about your implementation. It's literally just like a DSL between your implementation and the client-side. It's just saying that "For the API, here is a schema that works", and you can... |
**Jerod Santo:** You can munge it into that form however you want. |
**Coby Chapple:** Right. So that allows you to actually iterate on the implementation from a performance perspective in ways that you can't do with something where the implementation of the API is the implementation, if that makes sense. |
For example, one problem that we have a lot at GitHub is performance stuff. A lot of API queries... Someone asked for a bunch of information, but they might not actually need all of it. The response that we provide may include properties which are actually very expensive for us to compute... |
**Jerod Santo:** And they're gonna throw them away anyways... |
**Coby Chapple:** And they're not gonna use them anyway. So that's one of the huge benefits in terms of implementation - you can actually speed up your responses a huge amount if you're sure that they're only requesting information that they're actually gonna use, which actually allows you to provide something a lot mo... |
**Jerod Santo:** How could you cache anything ever though if it's always a very specific, customized response? |
**Coby Chapple:** I'm a designer, I don't know the specifics of how that works... |
**Jerod Santo:** You're speaking very fluently as a developer, so I just forgot that you're a designer... \[laughs\] |
**Coby Chapple:** \[31:59\] So I don't know from a cacheing perspective how that would work, but I think we're gonna see a lot more documentation and best practices start to become publicly available about how to do this stuff as we see more adoption. |
As we put our name out there and say "We're banking on this", I think a lot of other people are gonna start looking at it as an option, and I think that's gonna mean that people are gonna start talking a lot about the edge cases where it does get complex, like cacheing and performance... |
**Jerod Santo:** We got sidetracked on GraphQL... One last question, back to Projects, because I wanted to ask this earlier and I forgot to... You mentioned that you don't want to be like a project management tool, and as a designer you're probably very familiar with the idea of like-- not "keep it simple", because it ... |
**Coby Chapple:** Yeah. |
**Jerod Santo:** Why not just punt all the way and be like, "We don't wanna be a project management suite. Trello is great... We're just gonna tell people to go use Trello. Maybe we'll even provide some hooks to Trello..." Why not do that? |
**Coby Chapple:** That's a great question. My take on that is that just encouraging people to use third-party tools like Trello is an option for a lot of people. There's a lot of people though that would like to do that, but maybe it's not an option; maybe there's a security concern... Where maybe a larger company... |
**Jerod Santo:** So for your on-premise stuff... |
**Coby Chapple:** Yeah, how do we also control the access to this information? If we're suddenly then forcing people to also have Trello accounts, how do we make sure that the people seeing information are seeing information they should have access to? |
I think also Trello is a very general-purpose tool. We wanted to create something that's very basic, that's basic for people who build software - not just developers, but very basic for software project managers... There's a whole bunch of people around developers that also build software even if they're not directly c... |
As we iterate on this functionality and as we improve it, it's gonna keep coming back to, "Okay, what makes sense for people who build software?", and that's something that Trello has a much wider audience, so they may make other decisions that don't always make sense for people who are building software. |
**Jerod Santo:** Good answer. Okay, is there anything that I missed, like "I can't wait till he asks me this question" and I just never asked it? What else should I ask here? |
**Coby Chapple:** I don't think so. I don't have any questions at the top of mind that I... |
**Jerod Santo:** Do you have any questions for me? |
**Coby Chapple:** What surprized you the most about what you asked me so far? |
**Jerod Santo:** What surprised me about your responses? Good question... You should do this job. I mean, I think your take on Projects was refreshing -- the answer about on-prem was a surprise, because I had never looked at it from that perspective. I think I see GitHub very much from my lens, and my lens is a typical... |
**Coby Chapple:** \[36:10\] Cool. I think what that points to though is there's a lot of ways you can use GitHub, and what's gonna work for one company is gonna be completely different to what works for another organization, like a university or an open source organization. What works for different individuals and grou... |
**Jerod Santo:** Research. |
**Coby Chapple:** Right. So you're seeing this on github.com as well, where what works for one organization... |
**Jerod Santo:** Doesn't work for... |
**Coby Chapple:** Right, and that sure also happened at a smaller scale, too; within a repo, maybe that one repo might wanna use one project management plugin, and the other one just uses the vanilla one that we provide, and another one has an additional integration included in it, and that's all within one company. |
**Jerod Santo:** Right. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.