Comments on: I love pulling statistics out of git

By: MJ Ray

MJ Ray — Wed, 25 May 2011 11:11:26 +0000

What does Clay mean by NZ release? I’m pretty sure the release manager and maintainer of Koha 3.2 were both from the US.

LikeLike

By: Chris

Chris — Wed, 25 May 2011 01:24:14 +0000

Ok, I’m glad you are not angry.
I’m not sure where I made any judgement calls, and I measured both code bases from the same point in time until now. But if that offended your pedantry, I apologise.

I have rerun them from the commit you say is when you forked from Koha.
https://blog.bigballofwax.co.nz/2011/05/25/new-batch-of-statistics/

LikeLike

By: Clay Fouts

Clay Fouts — Wed, 25 May 2011 01:15:40 +0000

I’m not angry, just pedantic. If you’re going to measure something, it should be done accurately, and I knew your stats were wrong. The numbers don’t reflect much difference one way or the other, and “lines of code” or “commits” or “lines of deltas” aren’t informative data points about the progression of a code base, at any rate. Number of committers is somewhat more interesting, but only because having more eyes is nearly always better (though more hands is definitely not always better).

LikeLike

By: Chris

Chris — Tue, 24 May 2011 22:44:57 +0000

I picked that commit because that was the last version number,
3.01.00.061 that both had in common. Anyway to assuage your anger I have rerun them from the commit you suggested.
https://blog.bigballofwax.co.nz/2011/05/25/new-batch-of-statistics/

I never suggested you added 10,000 lines of ‘our’ commits, to your fork. In fact I don’t care about that, you are welcome to cherry pick as much as you like into your fork.

You are angry because you think I implied you cherry-picked something you didn’t? I’m not sure it matters, whether that commit was there before you forked or after, the commit is still in your repo. All I was interested in was the number of commits since the fork.

LikeLike

By: Clay Fouts

Clay Fouts — Tue, 24 May 2011 22:38:25 +0000

Yes, the date has nothing to do with it. I understand perfectly well how git works, which is why I used git-merge-base to determine the point of divergence rather than a date. What prompted you to use 58ee841 as a starting point besides the fact that allowed you to include 900k+ lines of translation file changes? It’s not 10,000 lines of of your commits that we’ve included in 4.2. It’s ten, as in the number after nine, which is the number of pluses and minuses in commit 45f2e56, the only one to have been picked in since the divergence.

LikeLike

By: Chris

Chris — Tue, 24 May 2011 21:57:24 +0000

I ran them both from 58ee841a73ea02d38d465e8c4663ab7bf509a62e to the head of the respective branches, that actually gives the impression that the Liblime fork has more commits on it. THanks for pointing that out, I will rerun it.

Also the date of the last commit has nothing to do with it, Koha has 3455 changesets, Liblime ILS 826 thats where the 10,000 lines come in. I think you are misunderstanding how git works.

I will rerun them from that commit you suggested to the latest release of both repos. Also what is this NZ release you speak of? Do you mean the Koha project release of 3.2.0 ?

Coincidentally the tool used to count this is gitdm, its what the linux kernel developers use, not something I have written. The numbers come direct from that.

LikeLike

By: Clay Fouts

Clay Fouts — Tue, 24 May 2011 19:36:35 +0000

I show the point of departure between the two repos being commit 3bab38c. You have exactly one commit more recent than that in the 4_02 branch of git://github.com/liblime/LibLime-Koha.git repo, and it’s not 985,443 lines long; it’s 10. So where exactly did you count the other 985,433 lines?

Incidentally, our 4.2 was released about the same time the NZ version’s 3.2 was, so that would be a more relevant point of comparison if you were interested in such a thing.

LikeLike

By: Chris Cormack

Chris Cormack — Tue, 24 May 2011 07:49:33 +0000

There probably is, but I don’t think it really matters, it’s interesting to me to see how much the translations changed. I think the important thing is the number of changesets and the number of developers anyway.

LikeLike

By: Paul Poulain

Paul Poulain — Tue, 24 May 2011 07:35:03 +0000

Chris, the glitch of counting translation commits makes those stats quite strange: Frederic has 2.7% of the changeset and 50% of the changed lines. Isn’t there a possibility to ignore changesets related to .po ?
Or add a small disclaimer at the changed lines stats, to explain those “strange” numbers.
(PS: don’t spend too much time on that, it’s very minor)

LikeLike