configuration files, RCS, CVS...

It is interesting how such issues are resolved in the M$-Windows world. On Fri, 09 Jul 04 09:24:30 +0400, Victor Wagner wrote: KF>>>> What happened here about storing configs in RCS? MG>>> $HOME/.* I hold CVS with my hands. KF>> Which one? And why CVS, is there RCS? VW> The person who taught me to work under *nix used to say that VW> the project has files, developers and machines. If any one of VW> three more than one - you should use CVS, not RCS. Um... Well, RCS quite allows multiple users to work, in a normal way. If there are several cars, then you can also get out. But I didn’t understand something about the files. This means that between files there may be some non-obvious relationship, which requires ensure synchronous versions of both files? I don’t understand well how RCS works, and CVS seems even more complicated to me, a confusing (for me) and overloaded with unnecessary functionality thing. Well, if we return to the configuration files, then why is RCS bad? You can, of course, compare it with CVS, or with other version control systems. But why? So far, nothing at all, no changes have been recorded. Why is there a tit in the sky if there is not even a crane in your hands? I think so ifjust write a simple wrapper over the editor to force I did ci -l for each file, it won’t be bad at all. If comments are allowed in the configuration file, then history changes ($Log$) can be seen right there. If several people they begin to change the configuration, even if only as one user, there is still some point - at least a comment will be written, and If you want, you can see what has changed. Then there if there really are several administrators are turned on all handles at the same time, you can list Send email changes to stay informed. I don’t see any obvious shortcomings in the option proposed above. Why not. No, I understand that someone needs CVS and what the hell else, but in my opinion this is far from a general case, but rather the opposite, narrowly specialized solution. The option with RCS, in my opinion, is acceptable in in most cases. From a working machine to a small server. I invite conference readers to express constructive opinions on this issue and, perhaps, develop some general recommendations to the approach to the issue of storing configuration files in the control system version, or otherwise recording the history of changes, or simply saving outdated configuration versions. Maybe someone is already using it similar organization of configuration management, it would be interesting to knowwhat kind, and how it is in general. It would also be interesting to apply all this to M$-Windows, and not just Unix-like systems (I assume you can export the registers to a text file, and process them already). Otherwise, last time the heated debate turned into a flame, but on the essence of the issue so nothing was said. VW> Accordingly, if you have dotfiles common to several machines, then in VW> CVS is the place for them. Maybe it's easier to copy them? There's no need to complicate everything so much necessary. The file has a modification date, you can synchronize it by date. The history of changes is not particularly needed.