Where is the history?

[famous]

Where is the history?

[avelis]

+1 to this. I am not sure if it is possible at this point but the biggest asset any version controlled code base will have is its versioned history. Much can shared be by it.

[LeonFedotov]

👍

[nitinhayaran]

👍

[unruthless]

👍

[andrewdeandrade]

When several projects that began as private internal projects have been made public, it is not uncommon for a repo to be squashed before publishing the repo publicly. For example, Angular, Meteor and React all have initial commits that were derived from an existing codebase.

meteor/meteor@d69c2d1
facebook/react@75897c2
angular/angular.js@c9c176a

Has the missing history from these three projects proven problematic for a reason of which we are not aware?

The original repo had dozens upon dozens of personal branches on it that we did not want to publish, each for a variety of different reasons. Furthermore, at this point with 150 public forks and many more clones out in the wild, I'm not even certain it is possible to put the history back onto this repo since this is a fresh start.

That being said we are listening and interested in hearing more. Is there something in particular you all are specifically interested in seeing? We're more than happy to answer any specific questions you may have about the code and why it is the way it is. Insofar as the commit history is concerned, I can't promise anything since there are discussions we need to have on this issue. However, if we do decide to retro-actively publish some of the history, can those who are intimately familiar with git suggest how this may be accomplished given all the clones and forks out there in the wild?

Nothing specific, it would just be interesting to see how the project was developed. Understand the need to squash the repository though.

[andrewdeandrade]

Unless anyone has any objections. I'm going to close this for now.

[tbranyen]

@malandrew ya it's fine to close. I was about to write a reply yesterday, but don't really have any justification other than being a little religious when it comes to Git. Seeing the history missing was a red flag. IMO it would have been nice to see condensed history at major landmarks.

I still plan on trying out Famous for a project and contributing if possible.

[andrewdeandrade]

@tbranyen I totally hear ya on this. For future internal projects we release, we'll investigate the complexity of trying out a condensed history approach. Do you have any examples or suggestions on how one might go about doing this or what constitutes a good condensed history?

One of the more difficult things we had to reconcile was licensing issues. For example, we ended up going with the MPL license, but in a few cases people erroneously applied a license to the copy when copying and pasting stuff and we didn't want some previous version to be available under a different license via the git history. For example, if you create a node.js module, the default license is ISC. if you go through and decide that isn't the license you want when you release then what do you do about the entire history of the package.json file? AFAIK, there isn't really a way to rewrite the entire history of a file to selectively remove a few lines. Is there?

There were a few more issues than just this. One example I can think of from my own work is a copy-paste job of a loader library from TJ Holowaychuk's component.io in one file. That library is MIT-licensed, but I never made any effort to cite it beyond a quick 2-3 line comment at the top stating where the code was extracted from because it was always meant to just stub out the functionality and I never intended to keep that loader library in my code. In fact, I'm replacing it with something completely different that shares nothing in common (code, design or interface) with the copy pasted code that was originally there*. In a case like this, when all this work was private and always intended to be temporary scaffolding that wouldn't be there anymore at launch, the history, when made public, ends up running afoul of the right way to document from where you borrowed things. It's not like there's an easy way to go back and give credit to that code in the past that no longer exists in the future.

Anyways, I'm a fan of your work and we've spoken a couple of times back on #backbonejs on freenode since before 0.3.3. Let me know if we can help any and feel free to join us on #famous . Are you still in the Boston area? If you're in the Bay Area ever, feel free to visit us. My email is [email protected].

• FWIW The original code was a simple CommonJS loader. I'm using AST manipulation and code generation to explicitly link things together in the correct order.

[AdrianRossouw]

ping: @tbranyen @avelis @LeonFedotov @nitinhayaran @unruthless @cameronwilby

I just wanted to point out that famous have changed their policies on this, and the entire commit history is now available in a new single merged repository (no submodules). They have also been moving to a bazaar style of development, where changes are being made in the open.

If this was one of the issues stopping you from using famo.us before, it might be a good time to re-evaluate that decision.

[avelis]

😎 👍