Has Linus Torvalds made literally 175 million commits?

[hafniz]

Describe the bug
@torvalds joined GitHub on September 3, 2011, and makes 2000+ contributions every year (as seen on his profile) that sums up to about 22k. However, your badge shows that he has a total of 175308.6k commits, which is far from reality. Even the number itself is absurd- one who commits once every minute would need 333 years to achieve this number!

Expected behavior
"Total commits" shows a number that is comparable to the number of contributions given by Github.

Screenshots / Live demo link (paste the github-readme-stats link as markdown image)

Additional context
None

[anuraghazra]

haha nice observation, lol.
Thanks for reporting this, i think the commits are just 17k and the kFormatter is failed to parse it properly maybe.

plot twist: linus is actually 333years old

[anuraghazra]

Okay so turns out, github returns this number from their server: 164882902 which should be correct since it's coming from their API.

and parsing it with a proper kFormatter results in "165M"

function nFormatter(num, digits) {
  var si = [
    { value: 1, symbol: "" },
    { value: 1E3, symbol: "k" },
    { value: 1E6, symbol: "M" },
    { value: 1E9, symbol: "G" },
    { value: 1E12, symbol: "T" },
    { value: 1E15, symbol: "P" },
    { value: 1E18, symbol: "E" }
  ];
  var rx = /\.0+$|(\.[0-9]*[1-9])0+$/;
  var i;
  for (i = si.length - 1; i > 0; i--) {
    if (num >= si[i].value) {
      break;
    }
  }
  return (num / si[i].value).toFixed(digits).replace(rx, "$1") + si[i].symbol;
}

nFormatter(164882902) // "165M"

[anuraghazra]

Damn then linus is actually 333years old

[hafniz]

Hmm that's weird... unless they have some bot that auto-commits to some secret repos that don't show up in the published contribution count. Even so those "secret" commits have made up 99.99% of the total 🤔

[iamabs2001]

bad! nothing correct https://github-readme-stats.vercel.app/api?username=iamabs2001

[Rongronggg9]

Okay so turns out, github returns this number from their server: 164882902 which should be correct since it's coming from their API.

Hmm that's weird... unless they have some bot that auto-commits to some secret repos that don't show up in the published contribution count. Even so those "secret" commits have made up 99.99% of the total thinking

The API counts commit in approximately this way: counts all commits from someone in all non-forked repositories, despite some of them having the same commit hashes. So the answer is clear, there are tons of non-forked Linux kernel source trees on GitHub, so each commit from Linus in each non-forked Linux kernel source tree participates in counting...

[Rongronggg9]

Randomly pick some Linux kernel contributors and you will find that their commits have also been counted so many times.

https://api.github.com/search/commits?q=author:tiwai

https://api.github.com/search/commits?q=author:jhedberg

https://api.github.com/search/commits?q=author:Icenowy

[hafniz]

The API counts commit in approximately this way: counts all commits from someone in all non-forked repositories, despite some of them having the same commit hashes. So the answer is clear, there are tons of non-forked Linux kernel source trees on GitHub, so each commit from Linus in each non-forked Linux kernel source tree participates in counting...

That tells a lot. Btw how did you know the way it counts?

[Rongronggg9]

The API counts commit in approximately this way: counts all commits from someone in all non-forked repositories, despite some of them having the same commit hashes. So the answer is clear, there are tons of non-forked Linux kernel source trees on GitHub, so each commit from Linus in each non-forked Linux kernel source tree participates in counting...

That tells a lot. Btw how did you know the way it counts?

Let us take a look at https://api.github.com/search/commits?q=author:torvalds and you will find the answer :)

[Rongronggg9]

The API counts commit in approximately this way: counts all commits from someone in all non-forked repositories, despite some of them having the same commit hashes. So the answer is clear, there are tons of non-forked Linux kernel source trees on GitHub, so each commit from Linus in each non-forked Linux kernel source tree participates in counting...

That tells a lot. Btw how did you know the way it counts?

Or you may directly check https://github.com/search?q=author:torvalds to get the same conclusion as mine.

[Rongronggg9]

#1691

current	my fork

context: #564 (comment)
The search API is entirely unreliable and we should never use it. Though using such a traversal could potentially cause rate limits, we should still get rid of using the search API. Why? Almost all Linux kernel contributors could never use this project as you have already seen above. It is absolutely unacceptable, isn't it? If we need to avoid being rate limited, we should consider using a database to cache users' commits in the past years.

Even I and @anuraghazra are affected:

current	my fork

If you want to check if you are affected, my deployment is https://github-readme-stats-rongronggg9.vercel.app/api