4 ms·

To set up my question, maybe I should share a sort of hazy crackpot vision or dream. So when I look at Git, I see a variety of postmodernism that's like “look,

To set up my question, maybe I should share a sort of hazy crackpot vision or dream. So when I look at Git, I see a variety of postmodernism that's like “look, we’re not being pretentiously intellectual or even pessimistic—we’re being realistic, software just is a postmodern world, Alice and Bob both have software that they call ‘Linux’ on their laptops, and they just aren’t the same source code, but they are both Linux, and trying to assess which one is the “true” Linux really does devolve into asking questions about power and basically in this case the power is vested ultimately in one king, Linus Torvalds, who says which one is more modern, and that’s fine as long as you understand that it’s arbitrary and anyone can fork the project at any time and create a rival power-structure and so forth.” Like, these ideas that I made fun of when I was a kid turn out to be amazingly practical things as finally revealed by git enlightenment. I would summarize the problems of various bugtrackers including Jira as being insufficiently “up with the times” therefore. They reflect the old modernist approach to truth. Schedules, deadlines, when was this “done”. But we know in software that really you want your source of truth of your bugtracker to basically be in the same Git repository as your code: one and the same commit needs to simultaneously close a bug as needs to properly change the source code to fix that bug, because the bug may be fixed in your “staging” codebase but that bugfix may not yet be deployed to “prod.” A reversion merged into prod as a hotfix needs to suddenly re-open the bug that we said was fixed-in-prod because now it no longer is. All of that. That's my hazy crackpot vision. :) Given all that, there is also a great difficulty with people using workflows like GitFlow which have the same “old way of doing things” misunderstandings. It’s not that those workflows aren’t good, they are good—for SVN repositories. You apply them to a Git repository and then you have this strange tension where the workflow and the VCS are working at cross-purposes sometimes. So it's also things like GitLab's CI/CD being a file checked into Git. This sounds great until you realize that nobody is expecting postmodernism to suddenly pop up here. “How did you merge that thing into prod when all of our admins were unreachable? Admins are supposed to have to push the button.” “Well, I was really in a bind, so I did something I would have never normally done: I pushed up my own branch `fix-master`, and in that branch I changed the CI/CD file to trigger deploys to production on pushes to `fix-master` instead of `master`.” But I had to!” Git was like “hegemony? what hegemony? if `master` has power then everyone has power.” Hah. So my question is, based on this description, I am not sure that I see you have “drunk the same Kool-aid” about Git, I am instead seeing a declaration that you want something fast-and-easy-and-supports-Git. Which, like, is fine. I don’t want to come across as crapping on your invention with a dream that I absolutely admit is hazy and crackpot, right? You have actually done the work and I am hypothetical and I have mad respect for you and your work. But I am sharing kind of my bigger vision to ask, “do you have a bigger vision? is it different than mine, some more systematic structure?” Is Tara just an easy way to do something like Jira with Git, or do you have a fundamentally new concept in mind that you are headed towards that will change how we track issues with our software?

wheresvic4 6y ago
I really enjoyed your setup to the question and it did really make me think about issue management in deeper detail. I think that one of the points that you raise is quite valid - that issue tracking should really be integrated into version control.
cat199 6y ago
holding 'truth' as a single fixed thing is not 'modernist' any more than being relativist is true because it is what people do in some social setting. you are talking about philosophical relativism and using a philosophically progressivist bias to claim legitimacy. at some point, Alice and Bob's linux will diverge to the point that Charlie will recognize neither as being linux, at which point there needs to be some basis for legitimacy to have a common ground, which in turn 'devolves into asking questions about power' (e.g. authority/legitimacy) one can certainly be too fixed in ones approach, but the inverse is also true. if bob, alice and charlie all work for an organization of people delivering 'the same linux' (whether open source/community driven or public company), what will a consumer of this 'linux' recieve when they get it? which in turn 'devolves into asking questions about power' .. ultimately we need to be on some same page, even if it is a bunch of post it notes on a desk and not a neatly bound hardcopy, so i wouldn't let the dualism implicit in being relativist create a false dichotomy whereby being organized at all is somehow outside of the realm of possibility. what does your vision of 'keeping track of being on the same page' look like, if the page doesn't even need to exist to be on it? or does so-called software 'enlightenment' require dissociating from all fixed concepts altogether? (show me the computer that will work this way and not be a magic spell, quantum included, since we are talking about 'being realistic')
- crdrost 6y ago
  I’m not sure that we should read postmodernists as being willy-nilly relativists. Like certainly the sort of “practical postmodernism” that I see in Git is not willing to say “Linux is BSD” the way a relativist might. Linux is a unix kernel and more broadly an operating system and a community led by Linus Torvalds for several years and a dozen other things besides, postmodernists would claim—and that there is no Archimedean “sturdy enough place to stand” to tell you which of these definitions is the “best”—but it is not a bouncy castle or a monad burrito tutorial or whatever nonsense I just “cook up”. Like the philiospical postmodernists’ criticism of the modernist “reason is going to solve the problem of finally accessing truth and we are gonna have a big truth party” is that reason essentially makes a 2D projection of a 3D truth, and so you have to approach the same truth from multiple perspectives with reason to get a full appreciation for it. Reason does not eliminate the problem of perspectives—sort of a Platonic fallacy, that reason will give you one single “world of the forms”—and that it shouldn’t even try because perspectives are great, let’s have some more perspectives. I’m tying this to the notion of Git’s branching model of course, each branch being something of a slightly different perspective or reference code for the same amorphous project of Linux. What consumers receive is just that then, a perspective on the broader project, one particular snapshot with various patches and so forth. In an OOP analogy Linux is a class and they receive an instance of that class, maybe. But yeah I think I am with you, dissociating from all fixed concepts altogether is probably not a great direction to go with this. You want some project organization and request or issue tracking. I am sorry if I sort of was suggesting that issue-tracking at all was meaningless, I definitely don’t agree with that. But I do think it instead has something to do with recognizing that the code to resolve an issue (if there is code at all, if it's not just getting closed as works-as-intended) leads to that issue having a “status” which needs to be tracked across multiple branches, and that this process is tedious and error-prone when we try to do it by hand.

wheresvic4 6y ago

I really enjoyed your setup to the question and it did really make me think about issue management in deeper detail. I think that one of the points that you raise is quite valid - that issue tracking should really be integrated into version control.

cat199 6y ago

holding 'truth' as a single fixed thing is not 'modernist' any more than being relativist is true because it is what people do in some social setting. you are talking about philosophical relativism and using a philosophically progressivist bias to claim legitimacy. at some point, Alice and Bob's linux will diverge to the point that Charlie will recognize neither as being linux, at which point there needs to be some basis for legitimacy to have a common ground, which in turn 'devolves into asking questions about power' (e.g. authority/legitimacy) one can certainly be too fixed in ones approach, but the inverse is also true. if bob, alice and charlie all work for an organization of people delivering 'the same linux' (whether open source/community driven or public company), what will a consumer of this 'linux' recieve when they get it? which in turn 'devolves into asking questions about power' .. ultimately we need to be on some same page, even if it is a bunch of post it notes on a desk and not a neatly bound hardcopy, so i wouldn't let the dualism implicit in being relativist create a false dichotomy whereby being organized at all is somehow outside of the realm of possibility. what does your vision of 'keeping track of being on the same page' look like, if the page doesn't even need to exist to be on it? or does so-called software 'enlightenment' require dissociating from all fixed concepts altogether? (show me the computer that will work this way and not be a magic spell, quantum included, since we are talking about 'being realistic')

crdrost 6y ago
I’m not sure that we should read postmodernists as being willy-nilly relativists. Like certainly the sort of “practical postmodernism” that I see in Git is not willing to say “Linux is BSD” the way a relativist might. Linux is a unix kernel and more broadly an operating system and a community led by Linus Torvalds for several years and a dozen other things besides, postmodernists would claim—and that there is no Archimedean “sturdy enough place to stand” to tell you which of these definitions is the “best”—but it is not a bouncy castle or a monad burrito tutorial or whatever nonsense I just “cook up”. Like the philiospical postmodernists’ criticism of the modernist “reason is going to solve the problem of finally accessing truth and we are gonna have a big truth party” is that reason essentially makes a 2D projection of a 3D truth, and so you have to approach the same truth from multiple perspectives with reason to get a full appreciation for it. Reason does not eliminate the problem of perspectives—sort of a Platonic fallacy, that reason will give you one single “world of the forms”—and that it shouldn’t even try because perspectives are great, let’s have some more perspectives. I’m tying this to the notion of Git’s branching model of course, each branch being something of a slightly different perspective or reference code for the same amorphous project of Linux. What consumers receive is just that then, a perspective on the broader project, one particular snapshot with various patches and so forth. In an OOP analogy Linux is a class and they receive an instance of that class, maybe. But yeah I think I am with you, dissociating from all fixed concepts altogether is probably not a great direction to go with this. You want some project organization and request or issue tracking. I am sorry if I sort of was suggesting that issue-tracking at all was meaningless, I definitely don’t agree with that. But I do think it instead has something to do with recognizing that the code to resolve an issue (if there is code at all, if it's not just getting closed as works-as-intended) leads to that issue having a “status” which needs to be tracked across multiple branches, and that this process is tedious and error-prone when we try to do it by hand.