Bug in GHCi

[Julian Seward (Intl Vendor)]

We fixed a bug last week which could have caused something
like this.  I'll have a look in more detail.

J


| -----Original Message-----
| From: Koen Claessen [mailto:koen at safelogic.se] 
| Sent: Monday, June 11, 2001 10:06 AM
| To: glasgow-haskell-bugs at haskell.org
| Subject: Bug in GHCi
| 
| 
| Hi there,
| 
| I came across a VERY STRANGE bug in GHCi. It is difficult to pin down.
| 
| I send a couple of modules. Running "main" in the module
| called Toggle, generates a file called "system.galf". This
| is wrong!
| 
| At line 14, it says "bool(true).". To generate this, it must 
| use definitions from the module "Theory.hs" that I have not 
| even used! In fact, it uses the definition of the function 
| "isBool" from "Theory.hs", that I do not use anywhere else.
| 
| It seems to somehow intertwine the definition of "bool" in 
| "Theory.hs" and the definition of "toggleSystem2" in 
| "Toggle.hs", as commenting out lines from "bool" will show.
| 
| Using Hugs, we can see that the correct behavior at line 14
| is: "toggle2(s(t0)).".
| 
| Sorry for the big files, but I could not pin this bug down
| in a good way. Feel free to request more information.
| 
| /Koen.
|