GHC 6.12 dependency checking many times slower than 6.10

[sunrayser]

Get feldspar-language-0.1 from hackage and build it:

 http://hackage.haskell.org/package/feldspar-language

 http://hackage.haskell.org/packages/archive/feldspar-language/0.1/feldspar-language-0.1.tar.gz

Performance:

"cabal configure; cabal cuild; time cabal build"

6.12:
real    0m24.316s
user    0m24.063s
sys     0m0.159s

6.10:
real    0m2.252s
user    0m2.021s
sys     0m0.231s

"time cabal build" with 6.10 after "cabal build"-ing it with 6.12 (ie no "rm -rf dist" before building with 6.10)
real    0m24.304s
user    0m24.094s
sys     0m0.139s

The last one is especially strange.

[simonpj]

Simon's looked into this:

Thu Dec 31 16:46:51 GMT 2009  Simon Marlow <marlowsd@gmail.com>
  * Rolling back: Make FastString thread-safe.
  Ignore-this: 8f21b256b0c86d167f8f6984d2b27a87
      
  This patch was the cause of the compile-time performance regression in
  #3796.  My guess is that it is due to the use of unsafePerformIO which
  traverses the stack up to the first update frame, and perhaps we have
  a deep stack when reading the dictionary from a .hi file.  In any
  case, since we're not relying on thread safety for FastStrings, I
  think the safest thing to do is back this out until we can investigate
  further.

I'm not sure if we can close the ticket. Simon?

S

[igloo]

Merged.