Procedure to make a SAM dataset (also called SAM project) from
a set of runs (say, all the good runs in a store, or all the
runs for a particular condition like solenoid-off runs for 
alignment studies) --

1)  Go to the SAM Dataset Definition editor at

http://d0db.fnal.gov/sam_project_editor/DatasetEditor.html

    [ I have made this a bookmark on the Captain's netscape personal 
      toolbar, but this is not guaranteed to be stable against bored
      captains playing with Netscape, of course.]

2)  Click on 'Create New' from the menu on the left-hand side.

3)  Fill in the 4 boxes in the top section as follows:

    Definition Name - pick a terse, reasonably descriptive name
                      avoid special characters
                      this will be how the dataset gets retrieved, so
                        keep it short and obvious, e.g. store547 
    Work Group - dzero
    Username - <your sam username>  [If you aren't registered with SAM,
       this will be a problem.  You can register from the web page
       http://d0db.fnal.gov/sam_admin/cgi/autoRegister.py .  Choose at
       least the dzero working group when you register.]
    Description - this is free form; again keep it terse and helpful, 
       but you can add details to augment the name, e.g. good runs from
        store 547

4) Use the Dimension Query box to define the data set:

   for example, if there were 3 good runs in store 547, use the
   query

   RUN_NUMBER 123966,123969,123977 and DATA_TIER raw

   Unfortunately, this IS case sensitive.

5) Add useful keywords in the white block under the words "Keyword Usages".
   You can transfer keywords over from the "Keywords" list or invent your
   own.  I propose we use the keywords "store-dataset" for a dataset composed 
   of all good runs from a given store, and "farm-dataset" for a dataset
   on which the offline farm should run the reconstruction. I gave both those
   keywords to my store547 dataset, so they are already in the list.

5) Click the Translate Constraints button to see, in the white window 
   below the button, exactly what files will be in the dataset, as you
   have defined it.  If the list of files looks correct, click the "Save"
   button to (gasp) save the dataset.   

If you need a more visual guide to all this, I saved a copy of the
Netscape page used to define the store547 dataset as the file sam.gif
in /home/d0cap - you can display it from a d0cap xterm with the command
display sam.gif.