Advice for "pseudo public" repository on a USB key for a single contributer project

   Hello Gitsers,

  I'm a Git newbee, and I have started a project, with a few month's of code
 in
 a local git repository.

 I am the only develloper/contributor to this repo. I have a clone of my
 repository
 on a USB key, for back up purposes. Once in a while I work on another
 machine
 where I have cloned a repo from the USB key.

 Since there are no servers involved, I have used pull command
 to move my 'HEAD' around :

 after working on machine1 I do :

   commit to machine1Repo
   machine1Repo  --pull--> USBKeyRepo

 when I switch on machine2 I start by bringing it up to date from the key :

   machine2Repo  <--pull-- USBKeyRepo

  and when I'm finished  :

  commit to machine2Repo
  machine1Repo  --pull--> USBKeyRepo

  So far I have done all of this using the master branch,
 and things don'g always go smoothly I sometimes have
 conflicts that don't seem logical, since I'm the only
 contrbutor, I should never need to do merges.

  From what I have read my USBKey repo is like a public repo,
 so I have tried using a bare repo, because since I never work
 directly on the usb key, the souces on this repo are just
 adding unnecessary complexity. So far I had no success,
 because the pull command doesn't recognize my bare repo,
 it seems that bare repos must me accessed via a daemon process.

  I suspect that I'm probably not using Git correctly (or optimally),
 which is why I am inquiring here.

 Should I be using a dedicated branch
 on each machine (as if I were two contributors) ?
 Is it wise to use pulls instead of pushes ?
 Should I be using a bare repo on my key ?

  Any insight would be appreciated.

 Thank you all

Hi,

2010/1/24 Maxime Lévesque <maxime.levesque@gmail.com>:

I think you mean "push", since what you want is to make the changes in machine1Repo available in USBKeyRepo.

I think you mean "push" here and s/machine1/machine2/ too, so that would read

  machine2Repo --push--> USBKeyRepo

When you make changes on machine2 and go back to machine1, you need to fetch/pull in your changes, just like you do for machine2Repo:

  machine1Repo  <--pull-- USBKeyRepo

What's your config?

Assuming you use master everywhere, I think you could use this in your config:

  [remote "USBKeyRepo"]
    url = /path/to/repo
    fetch = master
  [branch "master"]
    remote = USBKeyRepo
    merge = master

See above. You can't use one "instead of" another - they serve different purposes and don't replace one another.

Definitely, or else you would have a working tree in that repo too but it never gets updated - not to mention all the warnings git would show.

El 24/01/2010, a las 03:17, Tay Ray Chuan escribió:

For this kind of workflow I generally use pull in both directions. Push works only if it results in fast-forward merges in both directions, and if you are a little forgetful like me it's fairly easy to sooner or later forget to update one of the repos and end up making commits in both of them, leading to a situation in which at least one of the merges would have to be a non-fast-forward.

If you always pull, you are at least working in a non-bare repo at that point and you can choose to either (non-fast-forward) merge or rebase.

Unless you are really disciplined about synchronizing the repos before you ever start any work, then I think "pull" in both directions is the way to go.

Cheers, Wincent