Re: [Gnu-arch-users] Re: give us a hand with arch

[Dustin Sallings]

On Friday, Sep 26, 2003, at 03:43 US/Pacific, Stephen J. Turnbull wrote:

> What I'm working toward (slowly) is a model in which "arch commits"
> are almost as lightweight as a file save.  The additional effort
> involves a single keystroke selection from a small number of keywords
> (I have typofix, bugfix, comment, ontask in mind).  Then as the
> changes build up, arch starts to nag you to organize your commits.

	I agree, this is a good model to use when branches are as easy as they 
are.

	I'm really not bothered enough by the checkin model to have 
participated in this thread as long as I have.  I'm only questioning 
whether it enforces a particular type of user's preference on a user 
who does not have the same preference by erroring on a commit with no 
predefined log rather than prompting the user for a log message ala 
RCS/CVS/perforce/OpenCM/etc...

>     Dustin>  OpenCM does both in that you can edit your future
>     Dustin> checkin message at any time up to the checkin (at which
>     Dustin> point it will be presented to you for your final edit).
>
> Arch does, too.

	Well, this isn't an arch issue, it's a tla issue (not sure if that's 
what you meant here).  The difference I'm talking about is that OpenCM 
prompts you for a final edit at commit time, whether you've begun a log 
message or not.  I can see three modes that would be an excellent point 
for a user preference here:

        On commit:

                Always prompt for edit of log message
                Prompt for edit of log message if no log present
                Never prompt for edit of log message (current behavior)

--
Dustin Sallings