[cvsnt] Re: "cvs commit -r " problem

[Matt Schuckmann]

"Bo Berglund" <bo.berglund at system3r.se> wrote in message
news:42e0b84e.356221930 at news.cvsnt.org...
> On Fri, 22 Jul 2005 15:30:14 +0800,
> =?ISO-2022-JP?Q?=1B=24B8t=3F=2E0B=1B=28B=28Hsin-An_Hou=29?=
> <amistad.hou at gmail.com> wrote:
>
> >Hi all,
> >   i want to change the revision of some files. i use "cvs commit"
command=
> >=20
> >to change the revision of file as the following in WinCVS:
> >cvs commit -r 2.1 test.v
> >  =20
> >   then a text file is poped on the screen, and it shows the
information,=
> >=20
> >
>
>CVS: ----------------------------------------------------------------------
> >CVS: Enter Log.  Lines beginning with `CVS:' are removed automatically
> >CVS:=20
> >CVS: Committing in .
> >CVS:=20
> >CVS: Modified Files:
> >CVS:  Tag: 2.1
> >CVS: test.v=20
>
>CVS: ----------------------------------------------------------------------
> >
> >while i close the text window, the gui is suspended until i stop CVS=20
> >manually. after i stop CVS, an error message shows up
> >error message:=20
> >Log message unchanged or not specified
> >a)bort, c)ontinue, e)dit, !)reuse this message unchanged for remaining
dirs
> >Action: (continue) cvs [commit aborted]: received interrupt signal
>
> This means that you did not enter a log message...
>
> >and the revision of file isn't changed.=20
> >How to commit revision correctly?
> >i use WinCVS 2.0.2.3 (Build 3) & cvsnt 2.0.51.0
>
> The short answer is: don't!
>
> Revision numbers are *internal* cvs data to keep track of the changes
> of a file and have no meaning outside CVS itself. The only reason
> WinCvs exposes the revisions ia that they can tell you that a change
> has been done (by someone else) when you update a file from the
> repository.
> In future versions of CVSNT the revision numbers will all go away and
> the book-keeping will be done in another way. So don't mess with these
> numbers.
> Already the later versions of CVSNT do not accept the command with a
> forced revision number to stop this mis-use of the system.
>
>
> /Bo Berglund

I've heard people on this board say exactly what Bo said many times and it
bothers me.
While I agree that there is rarely (possibly never) any reason for a user to
require a specific revision number I don't agree that revision numbers are
strictly internal.
They are a way of refering to a historical version of the file, the most
common use of them might be for a code review in which the reviewee tells
the reviewers to look at specific revisions of specific files. If the
revision numbers are going to go away in the future how will we refer to
specific historical versions of a file?

Matt S.