RE: lines modified in newly-added files

[Jim.Hyslop]

Derek Price wrote:
> Neil Conway wrote:
> 
> | Jim.Hyslop wrote:
> |
> |> Maybe it could be turned on using a command-line option, e.g.:
> |>
> |> cvs commit -a
> |>
> |> -a Display the number of lines added/removed for new or deleted
> |> files.
> |
> |
> | I'm don't think that a flag to "commit" is the right place -- I
> | want this change to take effect retroactively on old repositories.
> | Also, a flag at commit-time seems too inflexible -- why not
> | postpone the decision about the format of the log until the user
> | actually asks for it, rather than hard-coding it at commit-time?
> |
> | I'd be fine with a flag to "log" if people think that's worth
> | doing.
> 
> 
> It seems to me that `log' would definitely be the place for any flag,
> especially as there should be no new information being saved to the
> RCS files, only an extra calculation during parsing of the RCS file
> and an extra calculation when the log information is printed out.
Yes, of course, I don't know what I was thinking when I wrote "commit".

> Of course, I'm not sure I see the utility of a flag here, regardless,
> as long as the change remains on feature.  The only useful script that
> may break that I can think of would be one that is calculating the
> same information Neil is proposing to add to `cvs log' and such a
> script could be easily modified to trust the new output from 
> CVS instead.
Yes, you're probably right about that.

> Jim, do you have a particular failure case in mind?  I am tempted to
> say that since this will only go into feature that a flag is not worth
> the trouble unless some specific problems are encountered.
No, nothing in particular. I was just trying to minimize the potential
impact of the change. On further reflection, it's probably not worth the
effort.

-- 
Jim Hyslop
Senior Software Designer
Leitch Technology International Inc. ( http://www.leitch.com )
Columnist, C/C++ Users Journal ( http://www.cuj.com/experts )