* Store jobset evaluations in the database explicitly. This includes

recording the builds that are part of a jobset evaluation.  We need
  this to be able to answer queries such as "return the latest NixOS
  ISO for which the installation test succeeded".  This wasn't previously
  possible because the database didn't record which builds of (say)
  the `isoMinimal' job and the `tests.installer.simple' job came from
  the same evaluation of the nixos:trunk jobset.

  Keeping a record of evaluations is also useful for logging purposes.

src/sql/hydra.sql

@@ -403,24 +403,47 @@ create table ReleaseMembers (
 );
 
 
--- This table is used to prevent repeated Nix expression evaluation
--- for the same set of inputs for a jobset.  In the scheduler, after
--- obtaining the current inputs for a jobset, we hash the inputs
--- together, and if the resulting hash already appears in this table,
--- we can skip the jobset.  Otherwise it's added to the table, and the
--- Nix expression for the jobset is evaluated.  The hash is computed
--- over the command-line arguments to hydra_eval_jobs.
-create table JobsetInputHashes (
+create table JobsetEvals (
+#ifdef POSTGRESQL
+    id            serial primary key not null,
+#else
+    id            integer primary key autoincrement not null,
+#endif
+
     project       text not null,
     jobset        text not null,
+    
+    timestamp     integer not null, -- when this entry was added
+    checkoutTime  integer not null, -- how long obtaining the inputs took (in seconds)
+    evalTime      integer not null, -- how long evaluation took (in seconds)
+
+    -- If 0, then the evaluation of this jobset did not cause any new
+    -- builds to be added to the database.  Otherwise, *all* the
+    -- builds resulting from the evaluation of the jobset (including
+    -- existing ones) can be found in the JobsetEvalMembers table.
+    hasNewBuilds  integer not null,
+
+    -- Used to prevent repeated Nix expression evaluation for the same
+    -- set of inputs for a jobset.  In the scheduler, after obtaining
+    -- the current inputs for a jobset, we hash the inputs together,
+    -- and if the resulting hash already appears in this table, we can
+    -- skip the jobset.  Otherwise we proceed.  The hash is computed
+    -- over the command-line arguments to hydra_eval_jobs.
     hash          text not null,
-    timestamp     integer not null,
-    primary key   (project, jobset, hash),
+
     foreign key   (project) references Projects(name) on delete cascade on update cascade,
     foreign key   (project, jobset) references Jobsets(project, name) on delete cascade on update cascade
 );
 
 
+create table JobsetEvalMembers (
+    eval          integer not null references JobsetEvals(id) on delete cascade,
+    build         integer not null references Builds(id) on delete cascade,
+    isNew         integer not null,
+    primary key   (eval, build)
+);
+
+
 create table UriRevMapper (
     baseuri       text not null,
     uri           text not null,