It's using SSH, port 22 typically, although that can be set as part of the
URL.
Gitosis is something for people like Github to run to manage mass amounts of
reopsitories, or maybe you could use it locally - git, and SCM becomes less
than half as useful if it's only on *your* machine, where is the safety net
if your laptop gets stolen.. your code is all gone too! [?]

- Lee

2009/2/5 Greg Hauptmann <[email protected]>

>
> Hi Lee,
>
> I see a blog post that had a Cap config entry like:
>
>    set :repository, "[email protected]:someuser/somegitrepository.git"
>
> which seemed to that capistrano at least is communicating to the
> repository over a network protocol.  So given there is no Git server
> what would the network layer typically be?  Like what protocol is
> Capistrano using when it communicates to the "github.com" address
> provided in the repository setting?
>
> Also I see Chapter 11 of "PragmaticVersionControl - with Git", is
> called "Running a Git Server with Gitosis".
>
>
>
> Regards
>
>
> 2009/2/5 Lee Hambley <[email protected]>:
> > There's no such thing as a "git server" -- and your web/db/app server
> needs
> > access to your repository.
> > To work like this, You'll need to use the "copy" strategy, which just
> > uploads a copy of the code to the server, or you may be able to tell your
> > server how to reach your laptop, but that is asking for trouble with NAT
> &
> > Firewalls.
> > My suggestion would be to buy a cheap, personal plan on something like
> > Github (or try one of the free providers) and experiment that way.
> > - Lee
> >
> > 2009/2/5 Greg Hauptmann <[email protected]>
> >>
> >> [ I'm up & running Git on my local dev Macbook :) ]
> >>
> >> Quick followup question - What's the simplest way for a hobbyist like
> >> myself to get capistrano deployment happening by Git?
> >>
> >> That is, whilst I have a local file Git repository on my Macbook, and
> >> noting that my target production server (so to speak) is a standalone
> >> Redhat linux box that I have, as the capistrano deploy code runs (on
> >> the prod server I believe) it will need to talk to a Git server via a
> >> network like address (e.g.
> >> [email protected]:someuser/somegitrepository.git) I assume.  So assuming
> >> there's no hidden Git process/server running on my macbook right now
> >> (a ps ax | grep git, doesn't reveal anything), what's the easiest way
> >> to get setup up?
> >>
> >> I assume the options may include the following?
> >> (a) install Git server on my local Macbook (which one?)
> >> (b) install Git server on my target Redhat prod box
> >> (c) just have a copy of the repository on my target Redhat server
> >> (i.e. file based) and reference this directly as this is where the
> >> capistrano scripts would be running from (not sure this is possible?)
> >>
> >> Also what sort of server is required?  (it's a given I need to set one
> >> up myself within my network).   I see mention of Gitosis?  Can I just
> >> front my existing working directory repository with a web server for
> >> http access (or is there a specific Git server that is required)?
> >>
> >> thanks
> >>
> >>
> >> 2009/1/30 David Masover <[email protected]>:
> >> >
> >> >
> >> > On Wed, Jan 28, 2009 at 8:43 PM, Greg Hauptmann
> >> > <[email protected]> wrote:
> >> >>
> >> >> thanks - Can I ask if it easy enough to set up my own central GitHub
> >> >> server? (like I do wiht svn)
> >> >
> >> > Yes. Very much so.
> >> >
> >> >> c) source backup <== which I'm talking about here
> >> >
> >> > I don't worry about that.
> >> >
> >> > The main property you have to remember is that EVERY checkout of Git
> is
> >> > a
> >> > full repository.
> >> >
> >> > At least one, probably several, of the Capistrano strategies using Git
> >> > will
> >> > leave a checkout on the server -- at least one.
> >> >
> >> > That means you have a remote repository, in that sense -- you can
> always
> >> > pull the latest deployed code from your webserver. Probably not a best
> >> > practice, but trivially easy.
> >> >
> >> > Or, just set up another user account, and push there. It's about two
> >> > commands to do that.
> >> >
> >> > Backup, well, if you do either of the above, you will always have at
> >> > least
> >> > two full copies -- the server, and your checkout.
> >> >
> >> > There's also always Github, though I'm not sure what the value is for
> a
> >> > one-man, private project. If it's open source, though, it's a great
> way
> >> > to
> >> > lower the barrier of entry -- one click for anyone with a Github
> account
> >> > to
> >> > fork it and start hacking.
> >> >
> >> > >
> >> >
> >>
> >>
> >>
> >> --
> >> Greg
> >> http://blog.gregnet.org/
> >>
> >>
> >
> >
> > >
> >
>
>
>
> --
> Greg
> http://blog.gregnet.org/
>
> >
>

--~--~---------~--~----~------------~-------~--~----~
To unsubscribe from this group, send email to 
[email protected]
For more options, visit this group at http://groups.google.com/group/capistrano
-~----------~----~----~----~------~----~------~--~---

<<inline: 33F.gif>>

Reply via email to