[cvsnt] Re: cvs edit without repository connection doesnt do local actions.

[Eliot]

EB: cvs edit failing doesn't stop me editing, it just makes it more inconvenient

>> IMO any option in the server side cvsrc should be *enforced* by the 
>> server.
>> I.e prevent client from commiting a change when a file is being 
>> reserved edited.
> 
> 
> That's way too late... what if you've just spent two days fixing a bug 
> and someone else reserved edited it?  Fine if it's text, but if it's a 
> binary build you're stuffed.

This is true, but in many cases one knows that the file is text, and/or nobody 
else is editing and/or there is no server side cvsrc.

I suppose I am emphasising that "Concurrent" should be the default behaviour 
when there is no info to the contrary.

Would it be possible to cache the servers CVSRC for use in case the server can't 
be contacted.

>> I can recompile (in theory anyway ;-) the client to ignore the servers 
>> cvsrc.
> If you want to produce a patch I'd consider it (make it an optional 
> switch as the current behaviour is too useful to be lost).

What is the "correct" behaviour of the -f switch (force)?
Right now as I read the code, if there is a server connection, -f will force an 
edit even if there are other editors.
But if there is no server connection it always fails.

    if(errors || (check_edited && editors_found))
vs this
    if(check_edited && (errors || editors_found))

Am I missing something obvious?

>> Or I am forced to 'go behind the back of' CVS and make my local files 
>> editable, losing the ability to unedit etc.  Nothing is gained by this 
>> result.
> 
> not really... if you really want it then write a batch file that does 
> the unedit for you - offline it's little more than copying the file from 
> CVS/Base into the working directory anyway.

Thanks for the idea. In wincvs I can write a Python script (after I learn 
python) to do what I want...

- Eliot