[darcs-users] Re: tag and release

[Eric S. Johansson]

Juliusz Chroboczek wrote:

> There's nothing preventing you from having multiple tags with the same
> name.  A user doing |darcs get --tag| will get the latest tag with
> that name; you'll need to use |darcs get --to-match| with a hash in
> order to get an older snapshot.
> 
> A better solution is to use tags such as
> 
>   hello-world-0.3-20041019
>   hello-world-0.3-alpha-3
>   hello-world-0.3-broken-12
> 
> and use a final version of
> 
>   hello-world-0.3
> 
> In that way, a user that does a |darcs get --tag=hello-world-0.3| will
> get the latest tag of 0.3.  In order to get 0.3-final and fail if it
> doesn't exist yet, you'd need to use --tag='^hello-world-0.3$'.

I didn't know I could use multiple types of the same name so I guess I 
could just keep tagging everything exactly the same until I change 
release numbers.  On the other hand, I'm probably just as likely to 
forget to change release numbers unless I build into the release process 
which means either form would work.

Although it easier multiple tag solution, would have me tagging things like:

Hello-world-0.3-forgot-)*&^(*%#@$#$@@#-template changes yet again
hello-world-0.3-forgot-MFING-documentation files
below-world-0.3-screw this I quit

that's roughly what my last few rounds of test and debug are frequently 
like.  The innumerable details one forgets because they were "just 
there" during testing are exceedingly frustrating.  I really need an 
environment where I can put in "don't forget this tags" exceedingly 
conveniently and then collect and clean them up over time.  Before 
anyone suggests using a simple Emacs buffer, my work is typically 
scattered over a minimum of three machines and more likely 7 to 10. 
With at least three of four different contexts overlaid on different 
groups of machines.

<way off topic> I'm now looking at one possible tool (keynote: 
http://www.tranglos.com/free/keynote.html) as a possible organizing aid. 
   problem being it's only on Windows for a single machine.  I would 
love something like IMAP for data and be able to access small 
repositories of data across the net.
</way off topic>

> I can't help wondering whether you shouldn't be branching instead of tagging.

I can think of a multiple different interpretations of how to do this. 
What are you thinking of?

---eric