[darcs-devel] Re: performance idea: a darcs db?

[Mark Stosberg]

On 2005-01-28, Juliusz Chroboczek <jch at pps.jussieu.fr> wrote:
>
> I also don't think that a relational database is a good idea.  Haskell
> likes coinductive datatypes.  SQL likes relations.  I don't think the
> two mix well.

I can't answer that directly, but I did check to see that there some
libraries already out there to interface Haskell to Sqlite. Here's one:

    http://haskelldb.sourceforge.net/api/

I was thinking about how you might quickly find which patches affect a
particular directory. That data could be stored in a table:

  CREATE TABLE file_patch_map (
      file_path           VARCHAR(5000),
      patch_hash          VARCHAR(100),
  );
  -- Each file/patch relationship can only exist one
  CREATE UNIQUE INDEX file_patch_idx ON file_patch_map(file_path,patch_hash);

Now finding the patches should be very fast:

 -- Find all patches that affect files beneath ./my/dir
 SELECT patch_hash 
    FROM file_patch_map 
    WHERE file_path LIKE './my/dir%';

If the Haskell API is anything like the Perl DBI, you should able to 
have the result turned in a Haskell array for easy handling. 

    Mark    

-- 
http://mark.stosberg.com/