[GHC] #15938: Hadrian's recompilation check is extremely slow

[GHC]

#15938: Hadrian's recompilation check is extremely slow
-------------------------------------+-------------------------------------
        Reporter:  bgamari           |                Owner:  (none)
            Type:  bug               |               Status:  new
        Priority:  high              |            Milestone:  8.8.1
       Component:  Build System      |              Version:  8.6.2
  (Hadrian)                          |
      Resolution:                    |             Keywords:
Operating System:  Unknown/Multiple  |         Architecture:
                                     |  Unknown/Multiple
 Type of failure:  None/Unknown      |            Test Case:
      Blocked By:                    |             Blocking:
 Related Tickets:                    |  Differential Rev(s):
       Wiki Page:                    |
-------------------------------------+-------------------------------------

Comment (by snowleopard):

 I think I know a way to reduce the number of file pattern matchings in
 Hadrian.

 Currently we generate build rules for each build context and there are
 hundreds of contexts, e.g. see generation of
 `Rules.Compile.compilePackage`:

 https://github.com/ghc/ghc/blob/master/hadrian/src/Rules.hs#L108

 Instead, we should generate just one build rule per logical target and
 parse various context parameters from the path, as implemented in
 `libraryRules`:

 https://github.com/ghc/ghc/blob/master/hadrian/src/Rules/Library.hs#L22

 We need to rewrite all build rules in the same manner. This will
 essentially replace hundred-long linear chains of file pattern matchings
 with a single matching and subsequent parsing.

 This is an obvious optimisation, but requires a good understanding of
 Hadrian internals. I am happy to volunteer to implement this, but I'm
 unlikely to find time until the end of teaching term in the university
 (end of December). If anyone has a chance to do this earlier, please go
 ahead, I can review a patch.

 P.S.: I still have no idea where the long pauses after `Ctrl-C`-ing come
 from.

-- 
Ticket URL: <http://ghc.haskell.org/trac/ghc/ticket/15938#comment:24>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler