Anders Andersson wrote:
This is a silly argument though, and even incorrect. Git is not github
or gitlab. One could easily have made a similar "cvshub" website.
Git is a lot easier to administrate as a single user than CVS is,
since the repo is completely self-contained in the project folder
while CVS needs its own infrastructure set up in order to get started.
However, with git it is also trivial to use local or remote servers if
necessary - without involving a "paid admin". I really don't see where
you're getting this from, or you have forgotten that CVS has a
separate repository that must be administrated.
Well, you can init a CVS repository to a local folder and you don't need
an administrator for that.
On the other hand, last time I was to a coding bootcamp, people could
not figure out how to set and run a project using Git because Git has so
many user-facing moving parts. This is more of a reflection of the
people at the bootcamp having pea brains, but it comes to show it is not
magically easy to manage.
Personally, I think CVS in the wild is easier to use than Git in the
wild because Git in the wild has been perverted.
If you want to contribute to a cvs project, you cvs checkout it, do your
changes, send a patch to the mailing list and get ignored. In Git world
you do end up dealing with the likes of Github and Gitlab, which aren't
Git except for the fact you WILL have to deal with them. That means you
need to register with their service, fork the project, git clone, make
your changes, generate a proper pull request, deliver it, then get
ignored. It is so many extra steps that are not intuitive.
And yes, I know Git does not have to be like that, but 95% of Git
activity IS like that, therefore I think complaining about it is not
missing the mark.
Also, the CVS servers you can deploy on OpenBSD use pledge() to good
effect and the whole thing feels a bit more solid than invoking
git-shell in the server. Maybe it is just me.
