> People say Linear is fast but it's nothing compared to how well Pivotal worked.
My company switched off Pivotal Tracker because it would slow to a crawl and require several seconds (!!) to load the page, with individual actions causing a DOM cascade that frequently hung browsers. Maybe it worked at small scale but it definitely didn’t work at a large scale.
That's weird. Pivotal itself had hundreds of developers using it for what was thousands of projects. It was never slow like that for us, unless they were having some sort of outage or something.
I hate to say it, but my immediate guess is that your company was using it in some fashion that it wasn't intended for. That was the real problem with the tool. Unless you actually worked at Pivotal, you never really got to learn how to use it the way they intended. The documentation was good, but nothing beats going to the source.
Oh so you mean mandatory pairing (which does away with the deep thinking required for some algos) and requiring "clean code" and other Uncle Bob BS doesn't contribute to actually scalable and efficient code?
As someone who spent a few years in the codebase for Tracker. There was some attempt at “Clean Code” but mostly basic DRY techniques. The product, itself, was plagued by being in a split-brain reality of Backbone and React (we were migrating as much as possible to React). We never addressed performance low-hanging fruit for large projects and it’s because it affected relatively few customers.
That said, pairing has its ups and downs. I think as other folks have said, you miss out on being able to go deep on problems that require ruminating. You also lack overarching reviews that help to keep architecture clean. The code might be clean, but the architecture might get really messy really fast.
I used to think the active pairing was just a fad or crazy talk. Then I went and worked at Pivotal and learned it first hand. The way they did it works.
Though for the times I paired, it does help for easy/medium problems, but for me it is absolutely a no-go for things you need to "sit and think about" (check docs, or keep more than two or three things in your head at the same time)
You learn to talk aloud instead of just doing everything in your head. Two minds are always better than one. It becomes a great collaboration.
This gets combined with how they "fire" people there. If your pair isn't keeping up and contributing enough, then people stop wanting to pair with them and they get rotated out to another team or entirely. That effectively means that the person sitting next to you, can actually help you.
It does get really intense because if you're having a hard day or week say in your personal life, you still have to be on your game or you have to explain your situation as best as you can. But the thing is, everyone who walked into the door of Pivotal, knew the rules ahead of time. You were there, because you wanted to be there, not because it was just another job.
I liken it a lot to joining the military, which is what it felt like. A bit cultish too. All-in-all, I had a great experience there, learned a ton, and it changed my thinking forever.
They spent A LOT of time building PT. It isn't something easy to replicate, especially if you didn't work at Pivotal Labs and get their knowledge of how to use it. But, you're right, I was just lamenting the other day that there is nothing like it. I wish Rob had just open sourced it.
My company (Singlebrook) built a PT workalike that works really well (in my humble opinion - ha). We're only using it internally and with our customers, but we use it every day and are really happy with it. It's not a feature-complete clone, and we don't intend for it to be, but we've probably built 70% of the features and have covered all of the basics of the project UI.
We haven't been planning to release it commercially, but if you're interested in trying it out, please reach out via singlebrook.com/contact and I'd be happy to set up a test account for you. If you happen to really like it, we can probably figure out an arrangment for you to use it.
Not technically a Tracker clone (we don't have points or all of its states) but we're in the neighborhood -- at StoryTime we've been writing about the development process monthly if any of ya'll are curious about what takes the most time in this domain.
Tracker's basics are pretty simple but getting the details right takes some finesse. Its multi-user update model is deceptively sophisticated -- being very responsive without ever losing data is a bit tricky, and you're probably making some tradeoffs you'd rather not make. The drag-and-drop behavior also took us a while to get exactly how we wanted it.
Close, but not PT. PT doesn't have the concept of status columns. It is effectively in the active running iteration, or it is in the backlog. What fits in the running iteration, is based on velocity.
I love that it is just GH issues under the covers though, that's smart.
Pivotal was far from a perfect company (if there can even be such a thing to begin with), but sadly, a lot of good things were lost in its latter days. This was one of them.
Are there particular attributes, behaviors, or properties you feel are important that you can't find elsewhere? I see you mentioned latency, for example — what else is key to you?
BTW that hero layout breaks in weird ways with Firefox. The heading text overlaps itself, the yellow note goes over the body text, the buttons have no bottom margin.
its tested in firefox and not pixel perfect but could not reproduce those issues. are you talking about ff mobile by any chance? could you share more details/screenshots and exact browser versions?
Librewolf 141.0.3-1 on Linux. I just went back to get a screenshot and now it's rendering ok. There was no real indication of missing CSS, perhaps a font failed to load last round and you're relying on the exact font proportions?
not really and if css did not load there should be way more issues than the ones described. anyways, would be thankful if you drop me a line on discord if you encounter this gain!
So far I have not found any of the attempted PT reboots inspiring. All of them are just rebuilding the PT interface, but nobody has innovated so far and done anything really new and different.
I think a lot of "Pivots" are still in bereavement over Pivotal and Pivotal Labs being gone.
I run https://bye-tracker.net and am also working on https://lanes.pm (still closed beta but open for more users soon). I just retested all of them and it really seems none are fully usable. I also see few improvements since April, it seems many of the teams either burned out trying to finish in time, were overwhelmed seeing what it really takes to make the details work after the fun parts are over or were disappointed seeing how low user numbers are from direct switchers.
I disagree with your 2010 theory, back then it would have been even less fruitful to make a viable alternative with the limited time and resources available. Especially as things like the UI state handling and some of the architecture was unheard of back then, years ahead of its time and even linear now does not fully match many aspects.
Similarly "Any half-skilled indie dev could have made a killing on making a good replacement" is just wrong,
a) you need to be a really great developer to make something that even nearly matches PT, there is more than is obvious to someone not deeply familiar with the internals of the product and 15 year of refinements that went into it. A few of the cloning teams were some of these indie developers who thought it would be fun and easy which was obviously very naive.
b) "Make a killing" is equally misleading. I would be surprised if its possible at all to break even on something like a pure PT clone. There are maybe 200 original users who are super loud and "in love" but this can easily mislead to thinking there are millions of users that would be immediately jump on and start paying. It takes probably 2 orders of magnitude more addressable market to make the financials work just by providing a PT clone (Which is why lanes.pm is focusing on making what tracker would need to be today not just catching up on what it was.)
Just give it a bit more time, there are at least 2 teams including lanes working on this and we all decided not to launch something half baked and buggy that loses your data.
Man, PT was so good. Linear is okay, but I find it really slow - more than once I've opened an existing issue and started editing it, only for a few seconds later the actual content to pop in and make a mess.
My biggest issue with Linear is that it doesn't think about cycles the way Pivotal did sprints. Or more accurately, it's in what PT called "commit mode" all the time.
If your cycle ends with a bunch of work undone, it rolls into the next cycle. If that cycle already a full capacity of work planned, well, now your cycle is at 150% or whatever. Repeat. The solution, presumably, is to stop creating new cycles.
But what PT would do, and seamlessly at that, is rebalance the sprints. Your project manager wants to move a number of stories to the top of the backlog? Great, but now that item projected for three sprints out is now four sprints out. The velocity and the projections were extracted from reality, not mandated from on high.
I asked Linear if they could add this -- a cycle rebalance. I was told, no, this simply isn't something they want to do, but maybe I could do it with zapier. Trying to get them to understand why I wanted this was like trying to explain air to a fish. It simply did not compute.
We're developing a do everything productivity app. We were using PT as our product management tool... 75 developers, 50 projects, 20k stories. We added PT functionality to our own app and migrated everything over in April. It works well now (we have a million hours of development time in the app so we understand what it takes to make something that is reliable and easy to use. Also had a daughter who was a PM at PL who taught me how to use PT.) There is more info and a beta sign up form at at getflo.com.
If you continue to not find what you need and are willing to be a subject matter expert on what Pivotal actually is (because I never saw it), I would be interested in building this. A lot of people share your sentiment so it could be successful, but it's hard to clone something unless you know the thing.
Please not yet another failed clone attempt. If you never used it, you have absolutely zero chance of replicating it, its something you need to have experienced.
My partner built Velocity Tracker [1], not a clone but it aims to implement the core philosophy of Pivotal. Any and all feedback would be most welcome.
What is your use case? (how many users, for what type of project?). Asking because tracking building a CRUD app is different than building an OS, a movie, a video game, building, etc..
> People say Linear is fast but it's nothing compared to how well Pivotal worked.
My company switched off Pivotal Tracker because it would slow to a crawl and require several seconds (!!) to load the page, with individual actions causing a DOM cascade that frequently hung browsers. Maybe it worked at small scale but it definitely didn’t work at a large scale.
That's weird. Pivotal itself had hundreds of developers using it for what was thousands of projects. It was never slow like that for us, unless they were having some sort of outage or something.
I hate to say it, but my immediate guess is that your company was using it in some fashion that it wasn't intended for. That was the real problem with the tool. Unless you actually worked at Pivotal, you never really got to learn how to use it the way they intended. The documentation was good, but nothing beats going to the source.
What is your best guess for why it got so slow?
Oh so you mean mandatory pairing (which does away with the deep thinking required for some algos) and requiring "clean code" and other Uncle Bob BS doesn't contribute to actually scalable and efficient code?
Color me shocked
As someone who spent a few years in the codebase for Tracker. There was some attempt at “Clean Code” but mostly basic DRY techniques. The product, itself, was plagued by being in a split-brain reality of Backbone and React (we were migrating as much as possible to React). We never addressed performance low-hanging fruit for large projects and it’s because it affected relatively few customers.
That said, pairing has its ups and downs. I think as other folks have said, you miss out on being able to go deep on problems that require ruminating. You also lack overarching reviews that help to keep architecture clean. The code might be clean, but the architecture might get really messy really fast.
I used to think the active pairing was just a fad or crazy talk. Then I went and worked at Pivotal and learned it first hand. The way they did it works.
I believe you.
Though for the times I paired, it does help for easy/medium problems, but for me it is absolutely a no-go for things you need to "sit and think about" (check docs, or keep more than two or three things in your head at the same time)
You learn to talk aloud instead of just doing everything in your head. Two minds are always better than one. It becomes a great collaboration.
This gets combined with how they "fire" people there. If your pair isn't keeping up and contributing enough, then people stop wanting to pair with them and they get rotated out to another team or entirely. That effectively means that the person sitting next to you, can actually help you.
It does get really intense because if you're having a hard day or week say in your personal life, you still have to be on your game or you have to explain your situation as best as you can. But the thing is, everyone who walked into the door of Pivotal, knew the rules ahead of time. You were there, because you wanted to be there, not because it was just another job.
I liken it a lot to joining the military, which is what it felt like. A bit cultish too. All-in-all, I had a great experience there, learned a ton, and it changed my thinking forever.
They spent A LOT of time building PT. It isn't something easy to replicate, especially if you didn't work at Pivotal Labs and get their knowledge of how to use it. But, you're right, I was just lamenting the other day that there is nothing like it. I wish Rob had just open sourced it.
My company (Singlebrook) built a PT workalike that works really well (in my humble opinion - ha). We're only using it internally and with our customers, but we use it every day and are really happy with it. It's not a feature-complete clone, and we don't intend for it to be, but we've probably built 70% of the features and have covered all of the basics of the project UI.
We haven't been planning to release it commercially, but if you're interested in trying it out, please reach out via singlebrook.com/contact and I'd be happy to set up a test account for you. If you happen to really like it, we can probably figure out an arrangment for you to use it.
It’s been a few years since I was at a company that has used it, but I very much enjoyed Shortcut (née Clubhouse).
Not technically a Tracker clone (we don't have points or all of its states) but we're in the neighborhood -- at StoryTime we've been writing about the development process monthly if any of ya'll are curious about what takes the most time in this domain.
https://buttondown.com/storytime-team/archive/
Tracker's basics are pretty simple but getting the details right takes some finesse. Its multi-user update model is deceptively sophisticated -- being very responsive without ever losing data is a bit tricky, and you're probably making some tradeoffs you'd rather not make. The drag-and-drop behavior also took us a while to get exactly how we wanted it.
We use Zube.io these days. Everything is just a GitHub ticket under the hood, it just slaps a PM friendly UI on them. I'm a fan.
Close, but not PT. PT doesn't have the concept of status columns. It is effectively in the active running iteration, or it is in the backlog. What fits in the running iteration, is based on velocity.
I love that it is just GH issues under the covers though, that's smart.
Pivotal was far from a perfect company (if there can even be such a thing to begin with), but sadly, a lot of good things were lost in its latter days. This was one of them.
Are there particular attributes, behaviors, or properties you feel are important that you can't find elsewhere? I see you mentioned latency, for example — what else is key to you?
We have a page dedicated to what was special. You can read the descriptions in the hero section here: bye-tracker.net
BTW that hero layout breaks in weird ways with Firefox. The heading text overlaps itself, the yellow note goes over the body text, the buttons have no bottom margin.
its tested in firefox and not pixel perfect but could not reproduce those issues. are you talking about ff mobile by any chance? could you share more details/screenshots and exact browser versions?
Librewolf 141.0.3-1 on Linux. I just went back to get a screenshot and now it's rendering ok. There was no real indication of missing CSS, perhaps a font failed to load last round and you're relying on the exact font proportions?
not really and if css did not load there should be way more issues than the ones described. anyways, would be thankful if you drop me a line on discord if you encounter this gain!
So far I have not found any of the attempted PT reboots inspiring. All of them are just rebuilding the PT interface, but nobody has innovated so far and done anything really new and different.
I think a lot of "Pivots" are still in bereavement over Pivotal and Pivotal Labs being gone.
I run https://bye-tracker.net and am also working on https://lanes.pm (still closed beta but open for more users soon). I just retested all of them and it really seems none are fully usable. I also see few improvements since April, it seems many of the teams either burned out trying to finish in time, were overwhelmed seeing what it really takes to make the details work after the fun parts are over or were disappointed seeing how low user numbers are from direct switchers.
I disagree with your 2010 theory, back then it would have been even less fruitful to make a viable alternative with the limited time and resources available. Especially as things like the UI state handling and some of the architecture was unheard of back then, years ahead of its time and even linear now does not fully match many aspects.
Similarly "Any half-skilled indie dev could have made a killing on making a good replacement" is just wrong, a) you need to be a really great developer to make something that even nearly matches PT, there is more than is obvious to someone not deeply familiar with the internals of the product and 15 year of refinements that went into it. A few of the cloning teams were some of these indie developers who thought it would be fun and easy which was obviously very naive.
b) "Make a killing" is equally misleading. I would be surprised if its possible at all to break even on something like a pure PT clone. There are maybe 200 original users who are super loud and "in love" but this can easily mislead to thinking there are millions of users that would be immediately jump on and start paying. It takes probably 2 orders of magnitude more addressable market to make the financials work just by providing a PT clone (Which is why lanes.pm is focusing on making what tracker would need to be today not just catching up on what it was.)
Just give it a bit more time, there are at least 2 teams including lanes working on this and we all decided not to launch something half baked and buggy that loses your data.
Man, PT was so good. Linear is okay, but I find it really slow - more than once I've opened an existing issue and started editing it, only for a few seconds later the actual content to pop in and make a mess.
My biggest issue with Linear is that it doesn't think about cycles the way Pivotal did sprints. Or more accurately, it's in what PT called "commit mode" all the time.
If your cycle ends with a bunch of work undone, it rolls into the next cycle. If that cycle already a full capacity of work planned, well, now your cycle is at 150% or whatever. Repeat. The solution, presumably, is to stop creating new cycles.
But what PT would do, and seamlessly at that, is rebalance the sprints. Your project manager wants to move a number of stories to the top of the backlog? Great, but now that item projected for three sprints out is now four sprints out. The velocity and the projections were extracted from reality, not mandated from on high.
I asked Linear if they could add this -- a cycle rebalance. I was told, no, this simply isn't something they want to do, but maybe I could do it with zapier. Trying to get them to understand why I wanted this was like trying to explain air to a fish. It simply did not compute.
We're developing a do everything productivity app. We were using PT as our product management tool... 75 developers, 50 projects, 20k stories. We added PT functionality to our own app and migrated everything over in April. It works well now (we have a million hours of development time in the app so we understand what it takes to make something that is reliable and easy to use. Also had a daughter who was a PM at PL who taught me how to use PT.) There is more info and a beta sign up form at at getflo.com.
We're using gitlab issue tracker and it's not even close.
JIRA and its interface is incredibly slow in comparison. An extra 30-60 seconds per task to create assign and organize
Seconds makes (billable) hours. I recommend JIRA for all my consulting work.
(I jest, I jest.)
If you continue to not find what you need and are willing to be a subject matter expert on what Pivotal actually is (because I never saw it), I would be interested in building this. A lot of people share your sentiment so it could be successful, but it's hard to clone something unless you know the thing.
Please not yet another failed clone attempt. If you never used it, you have absolutely zero chance of replicating it, its something you need to have experienced.
My partner built Velocity Tracker [1], not a clone but it aims to implement the core philosophy of Pivotal. Any and all feedback would be most welcome.
[1] https://app.velocitytracker.co/
1. Google Login, please.
2. Screenshots.
Google login has been added, screenshots coming soon—thank you for the feedback.
Awesome!
What was special about Pivotal Tracker?
What is your use case? (how many users, for what type of project?). Asking because tracking building a CRUD app is different than building an OS, a movie, a video game, building, etc..
I still like redmine
Redmine was pretty great.
CM42 Central is a workable replacement and FOSS but there's definitely room for improvement.
I really loved Pivotal Tracker but sadly a lot of people want to make everything so complicated.
How much would you be willing to pay for a replacement?
List of teams trying for reference -
"10 teams are racing to build a pivotal tracker replacement" -https://news.ycombinator.com/item?id=43378925
[dead]
Project management is one of the use cases we plan on excelling at Proma.ai
We are launching in 2 weeks.
I would love give a demo to anyone interested. Email is in profile.