Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Paris Metro lines on a Git graph

github.com

Post reply on HN
Print view View original

11–20 of 37 posts

Re: Paris Metro lines on a Git graph

#11

Post by _ikke_ »

a3n wrote: post #9

> I had to remove some interconnections, because there were some cycles. Because git doesn't allow cycles? Then as cool as this is (and it's very cool), perhaps this isn't an appropriate tool for this purpose. Or might the natural, human-built cycles be expressable as subrepositories?

The object model of git is described as a directed acyclic graph, thus by definition, it has no cycles, which wouldn't make sense for a version control system anyway.

That said, this is most likely just meant as a fun visualization, because the visualization does look like a railway map.

Re: Paris Metro lines on a Git graph

#12

Post by elsjaako »

a3n wrote: post #9

> I had to remove some interconnections, because there were some cycles. Because git doesn't allow cycles? Then as cool as this is (and it's very cool), perhaps this isn't an appropriate tool for this purpose. Or might the natural, human-built cycles be expressable as subrepositories?

I think the git file structure technically supports cycles, you just need to break sha-1 to do it.

I don't think the tooling is made for it either.

Re: Paris Metro lines on a Git graph

#14

Post by stinos »

a3n wrote: post #9

> I had to remove some interconnections, because there were some cycles. Because git doesn't allow cycles? Then as cool as this is (and it's very cool), perhaps this isn't an appropriate tool for this purpose. Or might the natural, human-built cycles be expressable as subrepositories?

> perhaps this isn't an appropriate tool for this purpose

Isn't the sole reason this is on HN, on the frontpage, exactly the fact that git and it's graphs aren't appropriate at all for drawing metrolines and yet somebody though of it and then did it as well?

Re: Paris Metro lines on a Git graph

#15

Post by wereHamster »

elsjaako wrote: post #12
a3n wrote: post #9

> I had to remove some interconnections, because there were some cycles. Because git doesn't allow cycles? Then as cool as this is (and it's very cool), perhaps this isn't an appropriate tool for this purpose. Or might the natural, human-built cycles be expressable as subrepositories?

I think the git file structure technically supports cycles, you just need to break sha-1 to do it. I don't think the tooling is made for it either.

The low level tooling can do that. When creating a commit you can pass arbitrary parents to the command. But as you said you'd need to break sha1.

> git commit-tree [(-p )...] < changelog

Re: Paris Metro lines on a Git graph

#16

Post by a3n »

scrollaway wrote: post #10
a3n wrote: post #9

> I had to remove some interconnections, because there were some cycles. Because git doesn't allow cycles? Then as cool as this is (and it's very cool), perhaps this isn't an appropriate tool for this purpose. Or might the natural, human-built cycles be expressable as subrepositories?

> perhaps this isn't an appropriate tool for this purpose Would Mercurial be more appropriate, or do you think DVCS are simply not an ideal way of mapping metro lines?

In response to your siblings, I think this use of git is delightful, and I don't want to take away from that. I love that people play with these tools.

As to Mercurial being better, I don't know but I doubt it.

My comment isn't even really directed at the appropriateness of DVCS specifically. It's just a general observation of using an acyclic tool to describe a network with cycles. We do that a lot, sometimes for play, sometimes for practicality, and sometimes because we just flip the fuckit bit. http://thechive.com/2009/11/11/there-i-fixed-it-23-photos/

Re: Paris Metro lines on a Git graph

#17

Post by myhnaccount108 »

scrollaway wrote: post #10
a3n wrote: post #9

> I had to remove some interconnections, because there were some cycles. Because git doesn't allow cycles? Then as cool as this is (and it's very cool), perhaps this isn't an appropriate tool for this purpose. Or might the natural, human-built cycles be expressable as subrepositories?

> perhaps this isn't an appropriate tool for this purpose Would Mercurial be more appropriate, or do you think DVCS are simply not an ideal way of mapping metro lines?

Better try Source Safe :)

Re: Paris Metro lines on a Git graph

#18

Post by misnome »

a3n wrote: post #9

> I had to remove some interconnections, because there were some cycles. Because git doesn't allow cycles? Then as cool as this is (and it's very cool), perhaps this isn't an appropriate tool for this purpose. Or might the natural, human-built cycles be expressable as subrepositories?

Well, you can do branches and multiple-parent merges, so as long as you don't care about directionality mapping to commits you can express the structure on connection, at least.

Re: Paris Metro lines on a Git graph

#19

Post by nightpool »

a3n wrote: post #16
scrollaway wrote: post #10

Earlier quoted context omitted.

> perhaps this isn't an appropriate tool for this purpose Would Mercurial be more appropriate, or do you think DVCS are simply not an ideal way of mapping metro lines?

In response to your siblings, I think this use of git is delightful, and I don't want to take away from that. I love that people play with these tools. As to Mercurial being better, I don't know but I doubt it. My comment isn't even really directed at the appropriateness of DVCS specifically. It's just a general observation of using an acyclic tool to describe a network with cycles. We do that a lot, sometimes for pl…

I'm pretty sure scrollaway's comment is sarcastic :D

Re: Paris Metro lines on a Git graph

#20

Post by scrollaway »

nightpool wrote: post #19
a3n wrote: post #16

Earlier quoted context omitted.

In response to your siblings, I think this use of git is delightful, and I don't want to take away from that. I love that people play with these tools. As to Mercurial being better, I don't know but I doubt it. My comment isn't even really directed at the appropriateness of DVCS specifically. It's just a general observation of using an acyclic tool to describe a network with cycles. We do that a lot, sometimes for pl…

I'm pretty sure scrollaway's comment is sarcastic :D

quite :)
Post reply on HN