Re: [Monotone-devel] where to put a (default) database? [Was: clone cleanup failure on Windows]

[William Uther]

On 03/03/2007, at 10:22 AM, Nathaniel Smith wrote:

> Hmm, odd, I seem to be in the minority :-).  I have a directory under
> ~/src/ for each project, so I have e.g. ~/src/monotone/, and then it
> contains my db in the top level, monotone.mtn, and a bunch of checkout
> directories on different branches, and build directories for those
> checkouts.  The nice thing about this approach is that it means all
> the data for a single project is stuck together in one place; this is
> sort of the model I have in the back of my head when I talk about how
> I think the UI should look.

I could modify clone so that it stored the database in "<working-copy- 
name>.db".  Or "<working-copy-name>.mtn-db".

e.g.

mtn clone venge.net -b net.venge.monotone

would produce a directory "net.venge.monotone" and a database  
"net.venge.monotone.mtn-db".

mtn clone -b net.venge.monotone venge.net monotone-src

would produce a directory "monotone-src" and a database "monotone- 
src.mtn-db".

If the database already exists it could be re-used.  If a database is  
specified, it would be used - as it is now.

Thoughts:

  - This would mean that the database could be kept around if the  
checkout later fails.
  - This would make the existence of the database more explicit.
  - It generates a file that the user didn't explicitly ask for.
  - This would massively simplify the code (because the db creating  
and population wouldn't need to be inserted into the middle of  
checkout code - you're really just doing three normal mtn ops in a row.)
  - It shows that mtn has another common use case where you want to  
run a series of commands in a row.  (Maybe I need to implement the  
suggestion on the UI wiki page where monotone logs the commands used  
with a timestamp in ~/.monotone/ somewhere and gathers it in  
occasionally.  Then we could use that as empirical data for UI design.)

Be well,

Will          :-}