#4900: DEPENDS pragma
  Reporter:  cdsmith           |          Owner:                  
      Type:  feature request   |         Status:  new             
  Priority:  normal            |      Milestone:  7.6.1           
 Component:  Compiler          |        Version:                  
Resolution:                    |       Keywords:                  
        Os:  Unknown/Multiple  |   Architecture:  Unknown/Multiple
   Failure:  None/Unknown      |     Difficulty:  Unknown         
  Testcase:  TH_Depends        |      Blockedby:                  
  Blocking:                    |        Related:                  

Comment(by parcs):

 I have attached a patch that implements this functionality in such a way
 that the `HPT` is not touched at all during the summarise stage. Instead,
 the required information that would have been read from the `HPT` during
 the summarise stage is now pulled out of the generated `ModIface` and into
 a field in the `ModSummary` during the upsweep stage. After the upsweep,
 the module graph is reset with the  updated `ModSummary`s containing the
 necessary, fresh data for use in the next load cycle. Overall, I think
 this implementation is nicer and easier to reason about than the previous

 (As a bonus, the `hsc_mod_graph` is now kept sorted in bottom..top order
 and so the "Ok, modules loaded: ..." message will list the modules in a
 logical order instead of the seemingly-random order they're currently
 listed. This change makes one of the tests in the ghci testsuite trivially
 fail, though.)

 I think the patch is incomplete, still, at least until the uses of
 `linkableTime` are looked over to make sure they don't operate under the
 assumption that `linkableTime` is simply the timestamp of the source file

Ticket URL: <http://hackage.haskell.org/trac/ghc/ticket/4900#comment:56>
GHC <http://www.haskell.org/ghc/>
The Glasgow Haskell Compiler

Glasgow-haskell-bugs mailing list

Reply via email to