{var%20f='http://v.t.sina.com.cn/share/share.php?appkey=1515056452',u=z||d.location,p=['&url=',e(u),'&title=',e(t||d.title),'&source=',e(r),'&sourceUrl=',e(l),'&content=',c||'gb2312','&pic=',e(p||'')].join('');function%20a(){if(!window.open([f,p].join(''),'mb',['toolbar=0,status=0,resizable=1,width=440,height=430,left=',(s.width-440)/2,',top=',(s.height-430)/2].join('')))u.href=[f,p].join('');};if(/Firefox/.test(navigator.userAgent))setTimeout(a,0);else%20a();})(screen,document,encodeURIComponent,'','','https://www.manongdao.com/data/attach/logo/logo.png', '推荐 Summer. ? 凉城 的问题《GIT vs. Perforce- Two VCS will enter… one will lea》','https://www.manongdao.com/q-254648.html','页面编码gb2312|utf-8默认gb2312'));){kind=link}
So I'm in the process of getting GIT sold at work. First thing I need is to convince everyone that GIT is better at what they're already used to doing. We currently use Perforce. Anybody else go through a similar sale? Any good links/advice?
One of the big wins is that we can work with it disconnected from the network. Another win IMO is the way adds/checkouts are handled. More points are welcome! Also we have about 10-20 devs total.
I think the one thing that I know GIT wins on is it's ability to "preserve line endings" on all files, whereas perforce seems to insist on translating them into either Unix, Dos/Windows or MacOS9 format ("\n", "\r\n" or "\r).
This is a real pain if you're writing Unix scripts in a Windows environment, or a mixed OS environment. It's not even possible to set the rule on a per-file-extension basis. For instance, it would convert .sh, .bash, .unix files to Unix format, and convert .ccp, .bat or .com files to Dos/Windows format.
In GIT (I'm not sure if that's default, an option or the only option) you can set it up to "preserve line endings". That means, you can manually change the line endings of a file, and then GIT will leave that format the way it is. This seems to me like the ideal way to do things, and I don't understand why this isn't an option with Perforce.
The only way you can achieve this behavior, is to mark the files as binary. As I see that, that would be a nasty hack to workaround a missing feature. Apart from being tedious to have to do on all scripts, etc, it would also probably break most diffs, etc.
The "solution" that we've settled for at the moment, is to run a sed command to remove all carriage returns from the scripts every time they're deployed to their Unix environment. This isn't ideal either, especially since some of them are deployed inside WAR files, and the sed line has to be run again when they're unpacked.
This is just something I think gives GIT a great advantage, and which I don't think has been mentioned above.
EDIT: After having been using Perforce for a bit longer, I'd like to add another couple of comments:
A) Something I really miss in Perforce is a clear and instance diff, including changed, removed and added files. This is available in GIT with the
git diffcommand, but in Perforce, files have to be checked out before their changes are recorded, and while you might have your main editors (like Eclipse) set up to automatically check files out when you edit them, you might sometimes edit files in other ways (notepad, unix commands, etc). And new files don't seem to be added automatically at all, even using Eclipse and p4eclipse, which can be rather annoying. So to find all changes, you have to run a "Diff against..." on the entire workspace, which first of all takes a while to run, and secondly includes all kind of irrelevant things unless you set up very complicated exclusion lists, which leads me to the next point.B) In GIT I find the .gitignore very simple and easy to manage, read and understand. However, the workspace ignore/exclude lists configurable in Perforce seem unwieldy and unnecessarily complex. I haven't been able to get any exclusions with wildcards working. I would like to do something like
To exclude all target folders within all projects inside Server/mainline. However, this doesn't seem to work like I would have expected, and I've ended up adding a line for every project like:
And similar lines for bin folders, .classpath and .projet files and more.
C) In Perforce there are the rather useful changelists. However, assume I make a group of changes, check them all and put them in a changelist, to then work on something else before submitting that changelist. If I later make a change to one of the files included in the first changelist, that file will still be in that changelist, and I can't just later submit the changelist assuming that it only contains the changes that I originally added (though it will be the same files). In GIT, if you add a file and them make further changes to it, those changes will not have been added (and would still show in a
git diffand you wouldn't be able to commit the file without first adding the new changes as well. Of course, this isn't usefull in the same way the changelist can be as you only have one set of added files, but in GIT you can just commit the changes, as that doesn't actually push them. You could them work on other changes before pushing them, but you wouldn't be able to push anything else that you add later, without pushing the former changes as well.I have no experience with Git, but I have with Mercurial which is also a distributed VCS. It depends on the project really, but in our case a distributed VCS suited the project as basically eliminated frequent broken builds.
I think it depends on the project really, as some are better suited towards a client-server VCS, and others towads a distributed one.
We have been using Git for sometime, recently our Git server's harddrive crashed and we could not revert back to the latest state. We managed to get back to few days old state. When the server was back up. Everyone in the team pulled/pushed their changes and voila, the server is back to current state.
I think in terms of keeping people happy during/ post switch over, one of things to get across early is just how private a local branch can be in Git, and how much freedom that gives them to make mistakes. Get them all to clone themselves a few private branches from the current code and then go wild in there, experimenting. Rename some files, check stuff in, merge things from another branch, rewind history, rebase one set of changes on top of another, and so on. Show how even their worst accidents locally have no consequences for their colleagues. What you want is a situation where developers feel safe, so they can learn faster (since Git has a steep learning curve that's important) and then eventually so that they're more effective as developers.
When you're trying to learn a centralised tool, obviously you will be worried about making some goof that causes problems for other users of the repository. The fear of embarrassment alone is enough to discourage people from experimenting. Even having a special "training" repository doesn't help, because inevitably developers will encounter a situation in the production system that they never saw during training, and so they're back to worrying.
But Git's distributed nature does away with this. You can try any experiment in a local branch, and if it goes horribly wrong, just throw the branch away and nobody needs to know. Since you can create a local branch of anything, you can replicate a problem you're seeing with the real live repository, yet have no danger of "breaking the build" or otherwise making a fool of yourself. You can check absolutely everything in, as soon as you've done it, no trying to batch work up into neat little packages. So not just the two major code changes you spent four hours on today, but also that build fix that you remembered half way through, and the spelling mistake in the documentation you spotted while explaining something to a colleague, and so on. And if the major changes are abandoned because the project is changing direction, you can cherry pick the build fix and the spelling mistake out of your branch and keep those with no hassle.
I have been using Perforce for a long time and recently I also started to use GIT. Here is my "objective" opinion:
Perforce features:
GIT features:
Overall for OpenSource/Distributed projects I would always recommend GIT, because it is more like a P2P application and everyone can participate in development. For example, I remember that when I was doing remote development with Perforce I was syncing 4GB Projects over 1Mbps link once in a week. Alot of time was simply wasted because of that. Also we needed set up VPN to do that.
If you have a small company and P4 server will be always up then I would say that Perforce is also a very good option.
What Perforce features are people using?
I ask because if all folks are doing is get and put from the command line, git has that covered, and so do all the other RTS.