Re: advise on the result on the this rcs scenario

[Aaron S. Hawley]

I'd go with Subversion if your colleague is bringing home a CD-rom, when
he'd be willing to check-out files from your source management over the
Internet. If you're thinking about CVS, also take a look at Subversion:

http://subversion.tigris.org/

Here's information on converting RCS to subversion:

http://svn.haxx.se/dev/archive-2002-10/2079.shtml

Tell your boss that Subversion and CVS were inspired by and are
backwards-compatible with RCS.

When your colleague takes home all the working files on CD, RCS was
specifically designed to take care of that scenario with the rcsmerge(1)
command:

rcsmerge -p -r2.8 -r2.9 f.html  > f.html.rcsmerged
# merges differences of version taken home on CD (2.8)
#                       version checked-in at the office (2.9)
#                       version brought by colleague from home (f.html)
# store changes in f.html.rcsmerged
#   merge conflicts are also put in this file, see rcsmerge(1) man pages.

you'll have to do a recursive diff on the directories to find which
changes were made by your colleague at home:

diff -lr /webroot/sites /cdrom/sites/
# list files that were changed

/a

On Fri, 14 Jan 2005, Mark Cooke wrote:

> Hi,
>
> Can someone please let me know if my interpretation of this is correct,
> my boss my set-up an rcs repository at work and after countless
> conversations, I feel that his needs would be better suited to using to
> cvs, rather than rcs.
>
> Here's the scenario:
>
> We have a central rcs repository, this is exported via a windows shared
> to each workstation, as the company doesn't want to install a web server
> on each machine and does not want every developer to checkout the files
> to a working directory on their local machine,
> they are making everyone use the same webroot (again exported via shares
> to each workstation) to develop on.
> This happens to be on the same server as the repository.
> So each person has had to checkout all the websites to this webroot
> share (yes they are all checking it out the the same location).
>
> Every developer must lock the file, do the changes then check the file
> back in, whilst this way of working is a little strange, it seems to work.
> So in theory we do not ever need to update the the working directory, as
> we are all working on the same file, in the same directory.
>
> Here's the dilemma, we now have one college works from home twice a
> week, he copies all the sites he will be working on to a cd, takes them
> home and does his changes on them.
> In the mean time we are busy back at work and may change some of the
> files he took home.
> We do the usual check then out, lock, edit, check in.
>
> Now here comes the problem, my boss is convinced that when my work
> college comes back on Monday, he can checkout the webroot version of the
> files he has been working on, and copy his files over the top, check
> them back in and it will commit his changes, *without* messing/loosing
> up changes.
>
> If I'm correct the only way to avoid disaster, is for his to check out
> and lock the webroot version, do a manual diff against his version and
> manually commit the changes, then check them back in, this will keep all
> the changes.
>
> Sorry if this sounds a bit muddled..
>
> All I'm realy after, if confirmation that my result and resolution (for
> this scenario) is correct.
> Yes I know this set-up is asking for disaster, but the boss set it up
> and will not alter it, but I feel that we should be using cvs instead.