Another small one. (Btw., I have just noticed that we still retain the two
"outer" iterations ... that may be subsumed by the adaptive support now, I
would say. Maybe add a low limit on number of iterations to 2 though, or
hardcode a single "warm up" run.)

Yours,
    Petr.

1 patch for repository http://code.haskell.org/darcs/darcs-benchmark:

Sat Feb 20 13:58:33 CET 2010  Petr Rockai <[email protected]>
  * Leverage the adaptive support in record benchmark as well.

New patches:

[Leverage the adaptive support in record benchmark as well.
Petr Rockai <[email protected]>**20100220125833
 Ignore-this: 41ebb9b6aa3f6c9ada5c315c1ca66c9
] hunk ./Standard.hs 75
   return ()
 
 -- | n patches for each file
-record_mod :: Int -> BenchmarkCmd ()
-record_mod n darcs _ = do
+record_mod :: BenchmarkCmd ()
+record_mod darcs _ = do
  cd "repo"
  files <- filterM test_f =<< ls "."
hunk ./Standard.hs 79
- forM_ [1..n] $ \x -> do
-   forM_ files $ \f -> liftIO (appendFile f (show n))
-   darcs [ "record", "--all", "-m", show x, "--no-test"]
- darcs [ "obliterate", "--last=" ++ show n, "--all" ]
+ forM_ files $ \f -> liftIO (appendFile f "x")
+ darcs [ "record", "--all", "-m", "test record", "--no-test"]
+ darcs [ "obliterate", "--last=1", "--all" ]
  return ()
 
 revert_mod :: BenchmarkCmd ()
hunk ./Standard.hs 109
        , Idempotent MilliSeconds "wh" wh
        , Idempotent MilliSeconds "wh mod" wh_mod
        , Idempotent MilliSeconds "wh -l" wh_l
-       , Idempotent MilliSeconds "record mod x10" $ record_mod 10
+       , Idempotent MilliSeconds "record mod" $ record_mod
        , Idempotent MilliSeconds "revert mod" revert_mod
        , Idempotent MilliSeconds "(un)revert mod" revert_unrevert
        ]

Context:

[Simplify benchmarks to account for adaptive testing.
Eric Kow <[email protected]>**20100220111523
 Ignore-this: e29f6262eacf2014da9ed6100c28193
] 
[Allow for benchmarks in different units.
Eric Kow <[email protected]>**20100220105406
 Ignore-this: 54146314f78d5059a4f532b8a8212bec
] 
[Back off from trying to implement Read on Benchmark.
Eric Kow <[email protected]>**20100220100445
 Ignore-this: a3a5336ef5586c2f516ea072ed63b89a
 Keep life simple.
] 
[Adaptive benchmarking to avoid ad-hoc (xN multipliers)
Eric Kow <[email protected]>**20100220095202
 Ignore-this: c850278f6c737794475ba69eff32faf4
] 
[Use authorspellings file.
Eric Kow <[email protected]>**20100220090957
 Ignore-this: 7a869a8ee495b4d5f15e44f074eff760
] 
[Warn that we're setting up a variant.
Eric Kow <[email protected]>**20100220002538
 Ignore-this: e07f5b9e4df44d48928f51c851fa4b33
] 
[Kill some trailing whitespace.
Eric Kow <[email protected]>**20100220002527
 Ignore-this: d51675a507684d7401ce474286d1767c
] 
[Give variants names that won't get picked up during darcs-benchmark run
Eric Kow <[email protected]>**20100220002445
 Ignore-this: e56446790e00f7e2125074a766124733
] 
[Resolve #3eb.
Eric Kow <[email protected]>**20100220000958
 Ignore-this: e5a8a4a46cee9c4b0e40e03bf199cfab
] 
[Bump to 0.1.6.
Eric Kow <[email protected]>**20100220000834
 Ignore-this: 29017313a3cb8c9502d8b55abf7c72f
] 
[Keep track of repo config files.
Eric Kow <[email protected]>**20100219234834
 Ignore-this: 82bd1195307bbbcea1e45b18b8268e63
] 
[Wibble.
Eric Kow <[email protected]>**20100219234200
 Ignore-this: 17f4aa956bbb8bab3027f00674c65bf0
] 
[Consolidate tables for repo variants.
Eric Kow <[email protected]>**20100219234133
 Ignore-this: 27ed470d50192764977c9528be515070
] 
[Simplify handling of variants.
Eric Kow <[email protected]>**20100219234111
 Ignore-this: f3921ce9dc1306baabd37e0ad97853e9
] 
[Do not use a dash in the variant suffixes.
Eric Kow <[email protected]>**20100219231020
 Ignore-this: 193fc83323a2b313beb5d550e7d94af7
 We want to abuse them for something else.
] 
[Add notion of "default" variant in case we ever add old-fashioned variant.
Eric Kow <[email protected]>**20100219230312
 Ignore-this: ae61928ec4935611309bba948c301782
] 
[Add a notion of "coreName" to TestRepo for tracking variants.
Eric Kow <[email protected]>**20100219223350
 Ignore-this: 520d7330cb6d616b7b6d8e78f8d514dd
 I'm getting a bit worried about how complicated this is starting to become.
] 
[Fix missing trailing blank line in RST table rendering.
Eric Kow <[email protected]>**20100219222949
 Ignore-this: 2d12ec4fbe2740b1b39105b4cd2a1f66
] 
[Run benchmarks on repository variants.
Eric Kow <[email protected]>**20100219222508
 Ignore-this: 59482af8df081b90789a207f7613c974
] 
[Create variants of repositories on demand.
Eric Kow <[email protected]>**20100219222506
 Ignore-this: 69a34812e91434e4c0a6aa54bbaf8767
] 
[Introduce notion of variants to config files.
Eric Kow <[email protected]>**20100219221237
 Ignore-this: 4c31a5c9792d1ee54cd46e188c24809e
 
 Examples of variants we may want to introduce
 
  - an optimize --pristine version of the repository
  - an old-fashioned one
  - in the future, a packed one
 
 The idea is to eventually support benchmarking of different variants to
 demonstrate the effect of eg, optimizing pristine.
] 
[TAG 0.1.5.1
Eric Kow <[email protected]>**20100216222241
 Ignore-this: 779d942891cee166e0119b09eba5abe9
] 
Patch bundle hash:
fe7308779a51da97999e361c332592524627425c
_______________________________________________
darcs-users mailing list
[email protected]
http://lists.osuosl.org/mailman/listinfo/darcs-users

Reply via email to