Trustworthy Systems

YOU ARE LOST in a maze of BitKeeper repositories — all almost the same


Peter Chubb

    School of Computer Science and Engineering
    Sydney 2052, Australia


Using BitKeeper to do source code management on the linux kernel for a platform not directly supported by the base operating system is hard. I describe the way I do it, using multiple repositories.

One has a local copy of the base tree (possibly at a slightly earlier revision than the head); a patched copy using third-party patches; a development copy (for generating clean patches to send to Linus); and a test copy (being a merge of the patched and development copies). It's important always to edit and commit in the development copy, then pull into the test copy for building and testing.

It can then become hard to work out where your changes actually are.

My conclusions are that source code control is a hard problem; distributed source code control is an even harder problem; and it's easy to get lost in a maze of BitKeeper repositories.

BibTeX Entry

    address          = {Sydney, Australia},
    author           = {Peter Chubb},
    booktitle        = {Australian Open Source Symposium},
    month            = jul,
    paperurl         = {},
    title            = {{YOU ARE LOST} in a maze of {BitKeeper} repositories --- All almost the same},
    year             = {2002}