[cvsnt] Re: cvs server: warning: new-born newfile.txt has disappeared

[Daniel Lapolla]

CVS will schedule it for addition in it's paret directory branch (In 
your example, probably trunk).

If wanto some other branch, you must do a cvs update (yes! in the file 
scheduled for addition - but be careful, cvs must know the branch 
already) to the branch, lets say br_yyy. And then commit it...



Bo Berglund wrote:
> On Fri, 16 Sep 2005 13:02:20 -0300, Daniel Lapolla
> <ldlapolla at yahoo.com> wrote:
> 
> 
>>>Can you describe exactly what you're doing?
>>
>>Imagine that developer 1, working on CR 12345 creates the branch 
>>br_cr_12345 on files A and B, and adds C directly in that branch.
> 
> 
> How is this done? If youi have a new file in a sandbox that is updated
> to HEAD then any file you add and commit will be added to TRUNK and
> not to the branch of the few files you actually have located on the
> branch.
> Imagine the following:
> 
> You have a sandbox on HEAD.
> Now you branch files 1, 2 and 3 to br_xxx and update these files to
> retrieve the branch. The remainder of the sandbox is at HEAD.
> 
> Then you branch files 4, 5 and 6 to br_yyy and update these files to
> retrieve the branch. Now you have 3 files on br_xxx, 3 files on br_yyy
> and the remainedr on HEAD.
> 
> Now you cvs add a new file.
> What is CVS supposed to do in this case? Add it to br_xxx or to br_yyy
> or to TRUNK? In my view CVS as only one way to know what to do and
> this is when you have updated the sandbox *directory* to the branch,
> then CVS reco´rds this in the CVS metadata and it will add the new
> file to the branch.
> 
> Is this what you are doing?
> If not then I don't see how you can get the new file to the branch at
> all.
> 
> 
> /Bo
> (Bo Berglund, developer in Sweden)