compiler crash under MacOS X

[Gregory Wright]

Hi Joerg,

On Aug 14, 2006, at 8:17 AM, Joerg van den Hoff wrote:

>
> gregory,
>
> thanks for your kind reply. sorry for responding so late but I was  
> OOO (Out Of Office).
>
>
>> Hi Joerg,
>> This problem is almost certainly bug #751.
>> This gives a compiler crash if the threaded run time system
>> is used.  The latest darwinports ghc (which I maintain) builds
>> the run time without threads.  What is the output of
>> sudo port info ghc
>> ?
>> If you do not have revision 2 of the 6.4.2 portfile, you need to
>> update:
>> sudo port selfupdate
> `port info ghc reported that it was _not_ revision 2 and so I did  
> the `selfupdate' as advised. this went OK. I then did
>
> sudo port upgrade ghc
>

Upgrading ghc (or any other port in darwinports) after you have had a  
build crash is not reliable.
There is no way to tell darwinports this yet, so you have to manually

sudo port uninstall ghc
sudo port clean --all ghc


And try to build again with

sudo port install ghc
>

> but this did not go so well:
>
> after some fetching/compiling it crashed and a second try yielded:
>
> --->  Activating ghc 6.4.2_0+darwin_8
> --->  Activating readline 5.0.005_0+darwin_8
> --->  Deactivating readline 5.0.005_0+darwin_8
> --->  Fetching readline
> --->  Attempting to fetch readline51-001 from ftp://ftp.cwru.edu/ 
> pub/bash/readline-5.1-patches/
> --->  Attempting to fetch readline51-001 from http://distfiles- 
> od.opendarwin.org/readline
> --->  Attempting to fetch readline51-001 from http://distfiles- 
> msn.opendarwin.org/readline
> --->  Attempting to fetch readline51-001 from http://distfiles- 
> bay13.opendarwin.org/readline
> --->  Attempting to fetch readline51-001 from http://distfiles- 
> od.opendarwin.org/
> --->  Attempting to fetch readline51-001 from http://distfiles- 
> msn.opendarwin.org/
> --->  Attempting to fetch readline51-001 from http://distfiles- 
> bay13.opendarwin.org/
> Error: Target com.apple.fetch returned: fetch failed
> --->  Deactivating ghc 6.4.2_0+darwin_8
> --->  Building ghc with target all
> Error: Target com.apple.build returned: shell command  
> "DYLD_FALLBACK_LIBRARY_PATH=/opt/local/lib make all" returned error 2
> Command output: make: *** No rule to make target `all'.  Stop.
>
>
> as in the first attempted `upgrade' the problem seemed to be the  
> unability to fetch readline51-001 from any site.
>


The problem is that you can't fetch readline, so that you can't build  
the required
readline dependency.  First, make sure your version of OS X is fully  
updated.
Apple has shipped broken version of curl, which is used by  
darwinports to fetch files.
Then make sure your DP is updated (yes, it is best to run

sudo port selfupdate

yet again.)

If you still have trouble a workaround was posted to the darwinports  
mailing list.

Note that if your DP installation is in /opt/local,  the fetched  
files are stored in
/opt/local/var/db/dport/distfiles/<port name>.  You can fetch the  
distribution
files manually and put them in the appropriate directory to work  
around such
problems.

Best Wishes,
Greg

> what is a poor guy to do? where is(are) my fault(s)? up to now I've  
> been 'fully elastically reflected' by ghc :-(
>
> all the best
> joerg
>
>> If you are running the latest version, your experiences
>> are... interesting.
>> Best Wishes,
>> Greg
>> On Jul 17, 2006, at 5:36 AM, Joerg van den Hoff wrote:
>>> hi,
>>>
>>> my very first tries with ghc, using the 'program'
>>>
>>> module Fact where
>>> fact :: Integer -> Integer
>>> fact n = product [1..n]
>>>
>>>
>>> gave the following compiler crash under MacOS X 10.4.7 (ghc  
>>> installed from source via DarwinPorts):
>>>
>>> Compiling Fact             ( fact.hs, interpreted )
>>> ghc-6.4.2: internal error: scavenge_stack: weird activation  
>>> record found on stack: 463
>>>     Please report this as a compiler bug.  See:
>>>     http://www.haskell.org/ghc/reportabug
>>>
>>>
>>> more irritating, this error was _not_ reproducible. a  second try  
>>> did run through:
>>>
>>> `ghc -c fact.hs' yielded a file `fact.o'
>>>
>>> but `ghc -o fact fact.o' yielded:
>>>
>>> /usr/bin/ld: Undefined symbols:
>>> _ZCMain_main_closure
>>> ___stginit_ZCMain
>>> collect2: ld returned 1 exit status
>>>
>>>
>>> and `ghc -v --make fact' yielded
>>>
>>>
>>> Glasgow Haskell Compiler, Version 6.4.2, for Haskell 98, compiled  
>>> by GHC version 6.4.2
>>> Using package config file: /opt/local/lib/ghc-6.4.2/package.conf
>>> Using package config file: /Users/vdh/.ghc/powerpc-darwin-6.4.2/ 
>>> package.conf
>>> Hsc static flags: -static
>>> *** Chasing dependencies:
>>> Chasing modules from: fact
>>> Stable modules:
>>> *** Compiling Fact             ( fact.hs, fact.o ):
>>> compile: input file fact.hs
>>> *** Checking old interface for Fact:
>>> Skipping  Fact             ( fact.hs, fact.o )
>>> *** Deleting temp files
>>> Deleting: /tmp/ghc18926.s
>>> Warning: deleting non-existent /tmp/ghc18926.s
>>> Upsweep completely successful.
>>> *** Deleting temp files
>>> Deleting:
>>> link(batch): upsweep (partially) failed OR
>>>    Main.main not exported; not linking.
>>> *** Deleting temp files
>>> Deleting:
>>>
>>>
>>> what I got where two files `fact.o' and `fact.hi' (attached).
>>>
>>> hope this helps in one way or the other
>>>
>>>
>>> all the best
>>>
>>> joerg van den hoff
>>>
>>>
>>>
>>>
>>>
>>> <fact.o>
>>> module Fact where
>>> fact :: Integer -> Integer
>>> fact n = product [1..n]
>>> <fact.hi>
>>> _______________________________________________
>>> Glasgow-haskell-bugs mailing list
>>> Glasgow-haskell-bugs at haskell.org
>>> http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs
>
> _______________________________________________
> Glasgow-haskell-bugs mailing list
> Glasgow-haskell-bugs at haskell.org
> http://www.haskell.org/mailman/listinfo/glasgow-haskell-bugs