- Date: Sun, 20 Jan 2019 21:12:21 -0800 (PST)
- From: Rusi Mody <rustompmody@xxxxxxxxx>
- Subject: Re: backintime
On Saturday, January 19, 2019 at 8:30:05 PM UTC+5:30, David wrote:
> On Sat, 19 Jan 2019 at 21:07, Andy Smith wrote:
> > On Fri, Jan 18, 2019 at 10:29:49AM +0000, Jonathan Dowland wrote:
> > > For those of you with decades of experience of CVS, you might as well
> > > stick with it.
> > >
> > > For someone entirely new to VCSes, I would absolutely not recommend
> > > CVS at all.
> > Yes. After reading the various diversions into RCS and CVS history I
> > was a little dismayed.
> Indeed, I also felt dismay at the idea that newcomers might follow advice to
> start using these ancient, incredibly limited tools.
> I'd be surprised if any of the people advocating them aren't well into
> retirement. I'm not trying to change their minds or opinions, I totally
> understand wanting to stay with the familiar, because that can be
> productive, and I absolutely agree with recommending the
> use of version control, but I feel that recommending RCS or CVS for
> new starters is extremely poor advice. The field of version control has
> seriously moved on from those early tools, which were widely abandoned
> and code migrated to more modern tools for legitimate reasons.
> It's not a fad.
> Here's a discussion of GIT features vs CVS ... it's ten years old.
Thats a good list of git-cvs comparison
One can get similar lists for svn vs cvs etc
What does that have to do with Gene's needs/request?
We can all agree with these facts
rcs followed by cvs followed by svn followed by git
>From which follows the conclusion:
git obsoletes svn obsoletes cvs obsoletes rcs
Except that the last 'obsoletes' is wrong because rcs is so much simpler
that it can be taken to solve a quite different problem altogether
To summarize the 1st 2nd 3rd version ideas from https://en.wikipedia.org/wiki/Revision_Control_System#Related_tools_and_successors
1st gen : file based revisions (rcs)
2nd gen : client-server model, concurrency (the 'c' in cvs)
Actually the first faltering steps towards
multi-user, multi-machine, multi-location multi-OS etc usage
(zillion other multis eg multi-line-ending support etc)
3rd gen : simplify client-server to peer2peer, disconnected usage, speed etc
What features beyond 1st-gen are of any use to someone with Gene's usage scenario viz. a single-user, single (config) file on a single machine??
Note that the fact that git is strongly biased towards projects (directories)
rather than files has made people have this kind of discussion
[see the accepted answer]
And even try to implement zit:
Note the blurb from https://git.wiki.kernel.org/index.php?title=Interfaces,_frontends,_and_tools#Zit
| Zit by Giuseppe Bilotta is the Git-based single file content tracker; it uses
| Git to independently track single files within a directory; sort of like what
| RCS does, but with the power, flexibility, elegance and ease of use of Git.
| Still in alpha stage.
| You can get it from `git://git.oblomov.eu/zit`