On Thu, 2014-11-06 at 09:18 +0000, Owen Smith wrote: > To give more detail, I meant create a branch, check the file out, test > your change, check change in to your own branch, then send details of > the branch to the master maintainers to see if they want to accept the > change (which seems to be a git pull request or similar), and if they > do they merge that branch back in.
Right. It's not a branch in the master repository, as Colin says, because people don't have access to that. But with a distributed version control system such as git, a 'branch' is fairly much interchangeable for most purposes with a completely separate clone of the repository elsewhere. You can happily push/pull code between them. > I wouldn't expect anyone, even dinkypumpkin, to do day to day work on > the top level master codestream. We don't. The master is out there on a server. We do the work on our *local* clone, and we test, and only then do we push it back out to the server. It isn't like CVS/SVN and other legacy systems where your "commit" goes directly to the server and is immediately visible to everyone. The distributed model doesn't work like that. You commit *locally* and it's only visible to the public when you push it to the server (or someone pulls it for you). > But patches seem so backwards. Source control systems do the job so > much better. For long sequences of patches, perhaps. But in many cases I *prefer* to receive a patch in email. After all, the first thing I see is usually an email anyway — even if it's just an email saying "please pull from...". So why not put the actual patch *in* the email, with its commit comment and everything in the standard layout. When I read the email it has everything *right* there. I can read the comment and the patch and make a decision, and if there's something I don't like then I can easily cite the 'offending' part of the patch in my email reply. And if I *do* want to apply it, it's trivial for me to save the email to a file and then apply it. That's much easier than having to click and bring up a web browser or something and go off looking at a remote git repository. Of course, all these things are subjective and some people will have different preferences. But that's why *I* still want to see patches in a lot of cases. -- dwmw2
smime.p7s
Description: S/MIME cryptographic signature
_______________________________________________ get_iplayer mailing list get_iplayer@lists.infradead.org http://lists.infradead.org/mailman/listinfo/get_iplayer