GitLab Plan Stage, 11 Jun 2019

Previous Meeting Next Meeting

⏯

youtube image

►

From YouTube: Conversation about GitLab's planning features

Description

Sean McGivern, Engineering Manager Plan Backend and Fabian Zimmer, Senior Product Manager Geo discuss some of GitLab's planning features

A

All right, wonderful yeah, so thank you. Thank you for taking the time today to talk a little bit with me about the the planning stage in in gitlab, so I think I managed, I use many of the features that that actually offers in that regard and I just went through sort of the twelve point. One kickoff document and I'm actually very excited about some of the things that are coming up so especially in plans for sort of all groups: the inheriting child children, epic start and due dates to parent ethics.

A

I think this is really cool and then also the creating issues from epics and creating child ethics from ethics. I think this is this is really nice, so I'm super you're actually getting to that awesome.

B

Yeah, it's funny right because of just how Planning works like because I'm, an engineering manager by the time we plan something I'm like oh yeah. That's basically I'm, not even really thinking about that right now, because it's like already like schedules like sure you thinking about the next thing. Yes, so.

A

What are you thinking about whether? What do you think is that it's, the next thing I mean.

B

I'm, not the product manager for plan the biggest issue I see at the moment for gitlab as an organization, but also for some of our we've heard this. We actually heard this initially from a customer and then we sort of stand off to talking about how it works. We get lab and I. Think Gio has this specifically as well is that everything works at the assumed for planning at the level of a top-level group. So you have your.

B

We have our gate lab or group which contains, like everything related to get out of the product, and you know you could have a label to that group. It will be available to any sub groups and sub projects within that group. You can have a milestone same thing. You can an ethic, you know you can add issues from any sub groups and so on and boards also work the same way.

B

The problem is, we also have a gate lab con group which has things to do with you of the company and that, because those don't share a top-level group, you can't use a board that touches both. You can't use an epic that touches both. You can't have the same label that touches both. There are some features and get loud together around that like if you filter issues on your dashboard by label it matches by title.

B

So if you have the same label title in two groups, you can see it there, but it's not unified, and you could say that the answer is just put everything in a gitlab group, but we also have like I, think we have a data group or something like that where the whole point of that group is that not everybody and get lab can see what's in there, and our permissions only go one way in groups which is that you're going to give people more access as you go down a group hierarchy, so that's I, think we're thinking about in terms of like how the market works.

B

I think that the term for these is like portfolios. So you have project management, which maps to get that projects program management which roughly maps to get that group. So there we have nested groups, so you could argue that we also have portfolio and in portfolio management is the next level up. So you could argue that portfolio management is possible because we have subgroups, but you could also argue that portfolio management will be a way of collecting things from different top-level groups, and we heard from a customer about this as well, because they were like.

B

We have a top-level group function in the organization essentially, and we want to be able to like do some of this stuff across all of them and putting everything in one top-level group seems really done, because the instance is just for us, so they have like the whole instance for us and then just like. Have one group in that instance like to make the features work, see my deal see how that's been on my mind, a lot, as you can probably tell I, don't actually have a great answer for how we do that.

B

But at the moment, what we're thinking is potentially have a portfolio feature or whatever that lets you aggregate top-level groups together and doesn't aggregate in terms of membership like it doesn't inherit membership, but it does inherit the plan features I mentioned ethics labels, boards milestones and then sort of go from there, like maybe in other areas like it will be useful to share stuff as well like obviously, merge request would be a good candidate, would just start with the plan stuff and see where it lands.

B

A

Know but I mean I'd like one thing that I I'm thinking about a little bit, so what I would I II at the moment you can.

A

You can already connect and say, like this issue is related to that issue and you can spend that potentially like. If you said I guess he said like within the group. That's fine right and then all of those things yeah.

B

So issue relationships actually works like maybe they just like work between any two issues like on the instance. As long as you have access to both of those issues, you can relate any two issues and we could do that for ethics as well. We could say, actually we're not going to enforce any structure here, we're just gonna, let you add any issue to any epoch, whether it's above below to the side of the epic, whatever I, think the reason we pick to do it. The more enforce way.

B

First, is that it's easier to go from stricter to more permissive than it is to then add on the.

A

B

If we'd started with that, and then we decided that actually it really made sense to have ethics and it might need to be enforced in this hierarchy that matches your groups. That would obviously be a nightmare like you know. How do you migrate that data like what do you do with ones already invalid? Like you know it's? Basically, it's not impossible, but it is a much bigger deal so.

A

B

A

Yeah I mean I have like also, and not quite sure, if this is the right forum, but I'll ask you anyways, so I have two things that are a little bit on my mind. That I think are interesting for, like a planning perspective, so I'm like I, used to use some other like product management like software's, where I could have more sort of semantic relationships between issues and epics and also like in a way to like visualize the dependency graph of things.

A

So rather than saying this is related to that which you know tells you that there is a relationship, but it just doesn't tell you why that they should exist or what that relationship signifies. So, for example, having the ability to say like this is a dependency of that right or like vice versa, or actually saying like hey. You know you need to this needs to be implemented. First, you know before you can do these other things, and actually it's like blocks they.

A

So this is yeah a dependency that we it's an interesting like and honestly I'm, not quite sure how how reasonable versus as I mean like I've used it I've liked it, but in a way in like humans, I think quite quite good at inferring, these relationships themselves, when you like, look at the links, but you always have to actually look at you know like what is this thing look at it. You know like and then kind of, establish it and opinions Mayberry, so I think that's a that's an interesting, an interesting thing. Yeah.

B

I think I think that is when we added that feature. We did explicitly talk about like adding types of relationships and future, and we just I think that was like two years ago, which is havin, haven't got round to it, like every cuz I remember when we were implementing it. We were talking about like internally like do. We call these two sides like a and B because they're arbitrary, because it's a relationship do we call them like source and target or something like that.

B

They imply some sort of relationship and we ended up going with internally there's something like source and target or, like you know from and to because if we add a relationship in future, then there will be like two sides of that relationship, whereas if it's just related like it doesn't really matter like you know, they could be either way around, but we did. We did have that in mind when we added it. I think the interesting thing about planning in general is like most of the stuff.

B

We add is a way of enforcing something you could already do, but you don't want to do it manually right. You know, instead of having an epoch, what we used to have were like messer issues where we collect past your issue, links and we'd say like okay.

B

Well, this needs to be done first, and this needs to be done afterwards and obviously like at the certain point that gets very painful to manage, and you can't introspect on that, like there's no API to that, like it's just a description, so I think yeah, it's a similar type of thing. We are um another thing, that's coming I, don't know when exactly because we had some issues with what sort of what we defined as an MVC for this, but like replacing the epic the with a tree.

B

I, don't remember that one things like so you yeah exactly all the ethics and issues within those ethics I represented in a tree instead of having to click through and see those. So that's the sort of start to visualizing that, but obviously that's one form or we can do that respond. I.

A

Think this is like I'd love that, because I think it's it's maybe I guess I have a like a slightly higher ethical, like structure in mind. Right I have like what I would like to do. What I mean this is. This is something where we're like I think moving in the right direction is like I I can write very high-level supervision ethics right that span. You know a large amount of time, and that should you know, convey you know like the general direction and I I think for like any stage or category.

A

He won't maybe like two or three of them, and that is something that you want to put on.

B

The roadmap and.

A

Said: okay: this is where we were going, but then you know within those are like some epics try that they start and they contain issues and I love to be able to like either see it in the tree view or to I. Have it actually on the ethnic roadmap where I can I can say like this? Is you know this? Is the roadmap I can see Beckett epics I can see all the issues that are contained within that ethic.

A

I can see the milestones you know at what these issues will be delivered and then I can a sickly create a roadmap that is as detailed or as high level as I want, depending on the audience. I can also sort of focus on a very specific part of this and like go into like into depth. Yeah.

B

That's really interesting, I don't actually know if we have an issue to do what you described on the roadmap, but I can definitely see how it would because, with the tree view, the road map is almost like a rotating of the tree view, so I'm, adding time and explicitly and I think I think that would be really interesting to be able to sort of visualize that, like you know, this is the span of the epic sound of child epic.

B

This is the span of the issues, but in the ethics based on their milestones, we have an issue for that. Yeah.

A

Like that would be good, because if we don't I tried to find it, but I didn't succeed, but it didn't look very hard. I can write a double.

B

I think sometimes people worried that, like you, create an issue, and that might already be one because, like we have like yeah so here print issues- and you know my perspective- is generally that's fine, like obviously you don't want to waste the effort you put into creating the issue but, like you know, put a reasonable amount of effort in, but not too much and that's fine. It's found ethic to view issues and children. Epochs in roadmap is currently on 12 top two. So there we go. That sounds like it is what we want.

B

Let me share that with you in the zoom chats but yeah. I think I think that does sort of match what you were talking about there, where you can expand a row and see like the detail of that road yeah. I think that would be very useful. I think.

A

That would be useful actually, and I think it's also interesting, because it would allow, for example, some of the like things that we do right now, like the kickoff calls, but to actually happen using a road map rather than a table way, basically go through things, and you say like hey for you. This is the release roadmap for that release, and it allows you to create interesting visualizations.

A

You know like focusing on a specific aspect, but on the release on an epic right on like different things, and I think thats- that's always really nice and it helps yes, keely yeah.

B

I think it's kind of interesting I think we sort of I mean with VSM as well, which is now part of managed, but like with all these things, we're sort of converging or like these tools that you can use to like arrange things in different ways like visually. It's like the same data but you're just exposing in a different way. So, like you, know, issues list an issue borders or it's a classic example.

B

For me, like you know, with an issue border, you can arrange things like horizontally and you might not arrange that for a workflow thing, like you know, I know: Eric who's. The product manager for plan at the moment uses a board that has milestone lists, so he and then it will filter that by like Direction issues or whatever, like you know, base all the plan issues. But then you know you can see like okay. This is my like, not exactly roadmap, but this is like my sort of short term like this.

B

Is this is where I'm going with these direction issues? This is this. Is the progress we're gonna make and then yeah the epics view? The roadmaps view is sort of a higher-level thing.

A

B

That data yeah.

A

I think this is I mean there's probably not one way like the one way to do. It, there's multiple ways to like slice like data in interesting ways, but I think from a sort of plan perspective I'm also from a competitive perspective. I think these are things that the entire, like software suites, that focus on like making load Maps desirable, rent for like presentation purposes, I think that is something where you know that's quite interesting and like just by, in my opinion, so also I. Think maybe technically.

A

This is a question for you, because I've worked with things where, as soon as I make a change in my roadmap and say like hey at these issues, I had to rerender the entire page, and it very quickly annoyed me and so I think, having sort of a dynamic way of like refreshing these, these visualizations and like adding things without having to leave and everything again, I, don't know how it is changing that this yeah.

B

I mean it's been, you know from some question, I guess so it's hard for me to unselect in too much detail. I can tell you that on a slightly rated note, the reason the epic tree may or may not be in 12.0 is because, when we added the initial version, like you know, I was talking about. What's the MVC here, so the MVC from an implementation perspective is to have it they read only but right now you can drag ethics and issues around to like reorder them.

B

So if we replace the view with one way, you can't do that then we've like regressed functionality, so even if from an implementation perspective, we start with read-only and then add the ability to drag things around. We still need to ship that as one thing to customers, we did sorry to users. We did talk a little bit about like having the option of the two views, but the problem was because we're gonna get rid of the old view. Shortly anyway, it seemed dumb to like add the option.

B

Flight would probably be one release and then remove it. We'd, rather just like you know, put it behind the feature flag and then make it available when it's ready, but yeah. That was her other thing that actually stopped that, because Europe will be able to draw an issue through the hierarchy, not just within its own epic. One sits in a tree view like you'd, expect so you'll be able to drag. Thank you from a parent epic to a child, epic, or vice versa, or to a sibling, epic, as well as reordering it, which.

A

B

But it's more work.

A

Specially most most interesting things tend to be. You know.

A

I mean just actually having child epics and being able to render a hierarchy or a tree. This is really interesting and.

A

B

So as a product manager like what's the kind someone whose use get live quite a lot, I think in the past, like what's the biggest thing that you feel like you missed from other tools like it might not be you can. You could say something: that's not like feasible for us to do soon. Like you know, that's fine, I'm, just curious about like what's the biggest paid point, you feel when using get that for this stuff. So.

A

I actually like one of the biggest pain points like when we had to make a decision in my in my previous previous life on what to use was that pigs were not available at that time, and so that is something where I was just like. I can't I can't handle that you know it's a. It makes no sense to me, I. Think at the moment, like so a couple of things we talked about the motorhomes I wish. We had like a more sort of complete roadmap experience and I.

A

Think both maps are such with powerful way to convey to like multiple audiences, what you are, after at different levels, I think for me, as a product manager, I'd love to be able to produce a road map for engineers and like people that, like I, feel like this is what we're going to do. Look at it right visually I'd, love to present a road map that is compelling to customers at a level where they really understand the value that we are with bringing and then like a very sort of high level where we going.

A

You know what is what's happening. The next three months, so that's that's something, that's cool, I, think and that's maybe a sort of a product manager like issue I, don't know right.

B

But like the plan stages, like maybe.

A

B

A product I just right, yeah.

A

I think the one thing that I think we don't do at the moment that I wish we did better and that confused me a lot in the beginning is we we get lab, we use. Okay, ours, don't electives and key results and I. Think that really powerful and I I'm used to I was used to sort of always asking myself when I would create an epic or you know, even at an issue level to be able to say like okay. Why are we doing this?

A

You know like what objective are we actually like driving forward here and I? Think at the moment, we don't really have a way to say, like this thing read relates to one of the like objectives that we are driving forward with those key results and I. Think you can do that by labeling right and by by doing it in that way, so I think that obviously ways of handling it, you can create issues that are okay, ours and linked them, but I I think having an explicit layer that handles okrs, pretty cool.

A

In my view, yeah.

B

No I think that's interesting. It's funny when you mention labels, there was like the advantage and the problem with labels is that you can use them for so many things and I think they're quite again, sort of going back to earlier they're quite unstructured, and if you want a more explicit structure it's hard to.

B

We do have some ways of creating that right now we have scoped labels, they're mutually exclusive, but like fundamentally it's quite hard to enforce a structure using something as free as labels, which is great as well, because it's very powerful and tools.

A

B

A lot of, but it's exactly, they also share the same visual space, I.

A

Think a big level so if I say like this is the this is an epic in in my in my right mind an epic, usually groups, a bunch issues that environment teachers and I, like I, always said like okay, if all the issues that are like in an epic I actually implemented and the epic should ideally be closed. Yes, so if that's that's true, so what I would love is sort of on an epic level information saying like this epic contributes in a measurable way to work.

A

You result, you know that is relevant rune objectives and making that really really clear and I. Think having that functionality, integrated I think would be quite cool. Yes,.

B

I mean another way you could do it at the moment is create epics for the objectives and then have sub epics for those, but again there's nothing. Stopping you from having an epic, that's, not a child of an objective. You know, there's nothing, there's nothing! Stopping you from just like crazy and I pick at the top level. Essentially and saying like you know, this is my epic we're gonna work on it. It doesn't relate to any of the objectives, but I don't care we're gonna work on it, which yeah so yeah, no I.

B

Think that's interesting. That's.

A

Another one where I, actually it's on my list to follow up if that is like written down somewhere I will do that. Yeah.

B

I'm definitely not aware of something like that, but eirick the product manager might know more about yeah.

A

We had like a tiny discussion and I think I think I mean okay, yeah, so I'll. Add that as one of my action items from from.

B

Just curious about how you use things, what you're excited about what you feel like it's missing and yeah. It was great to have the chat anything else you wanted to bring up before we wrap up yeah.

A

I think the only the only thing that I wanted to say is that it's amazing and wonderful that you took the time. So thanks very much I know you're quite well as.

B

You saw from my calendar today really have much of today in terms of beating, so as that.

A

That's true, but it just means that you will do other things so I think this is, you know I'm, not assuming that, because you have no meetings, you are, you know like sipping, you know like a nice drink in your living room, so yeah.

B

No, that's true, no yeah, no problem. It was great to speak to you as well and yeah I. Guess you I'll speak to you around. If you upload this to YouTube afterwards, let me know and I'll add it to the fun playlists and whatnot, but yeah.

A

So that it works so yeah I.

B

Could talk you to her on slack if you need help? Thank you have a good one.