MUST RE-FORK REPOSITORY (Large files in history)

[andyleejordan]

I noticed when cloning a superproject that uses DSC that, while the working tree is small, the repository itself is massive. I ran a quick script to identify the files, and it looks like we have some large tarballs in the history of ext:

69481257 ext/boost/boost_1_55_0.tar.gz
16843278 ext/python/Python-3.3.4.tgz
14725931 ext/python/Python-2.7.6.tgz
13318547 ext/python/Python-2.6.6.tgz
12787688 ext/python/Python-3.2.3.tgz
11019675 ext/python/Python-2.5.tgz
9239975 ext/python/Python-2.4.2.tgz

My feeling is that we really don't want Boost and Python distributions in our history. While they've been removed at some point, we likely want to run a filter-branch to purge them (unfortunately rewriting master again).

@jeffaco What are your thoughts on this?

[jeffaco]

I tend to agree. The fact that they were removed to begin with is an indication we do not need them.

Note that other (rather massive) changes are happening. SCXcore will have it's history rewritten to remove some 100MB, and a fair number of repositories are moving to the Microsoft group. In the end, to recover from this, I was planning on hitting the build/Jenkins systems and just deleting the repo directories manually, forcing a fresh checkout on the next build. Also, it's easiest for all the developers to just re-clone the superproject after this happens.

Shall I trim the DSC project at the same time?

It's unfortunate that Issue #23 forced users to create new forks too, and this was just a week ago ...

@Microsoft/ostc-devs
@Microsoft/omsagent-devs

[jeffaco]

Okay, history is rewritten. NOTE: Anybody that has forked this repository will need to fork it again (sorry)!

Old repository size: 228 Megabytes.
New repository size: 29 Magabytes.

@andschwa

[andyleejordan]

Very nice. Thank you @jeffaco! I've updated our project. Feel free to close.

[jeffaco]

Thanks for pointing out this issue!