Andreas Abel | 18 Oct 15:04 2013
Picon

Horray, a ghc bug!

You do not get them often, but here is one...

[208 of 254] Compiling Agda.TypeChecking.MetaVars.Occurs ( 
src/full/Agda/TypeChecking/MetaVars/Occurs.hs, 
dist/build/Agda/TypeChecking/MetaVars/Occurs.o )
ghc: panic! (the 'impossible' happened)
   (GHC version 7.4.1 for x86_64-unknown-linux):
	compiler/rename/RnSource.lhs:429:14-81: Irrefutable pattern failed for 
pattern Data.Maybe.Just (inst_tyvars,
                                                         _,
                                                         SrcLoc.L _ cls,
                                                         _)

Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug

--

-- 
Andreas Abel  <><      Du bist der geliebte Mensch.

Theoretical Computer Science, University of Munich
Oettingenstr. 67, D-80538 Munich, GERMANY

andreas.abel <at> ifi.lmu.de
http://www2.tcs.ifi.lmu.de/~abel/
Stijn van Drongelen | 18 Oct 15:15 2013
Picon

Re: Horray, a ghc bug!

Hi,

Have you tried the same with 7.6.1? When I search for RnSource.lhs:429 on Google, I find multiple GHC tickets, marked as 'solved in 7.6.1'.

Kind regards,

Stijn van Drongelen


On Fri, Oct 18, 2013 at 3:04 PM, Andreas Abel <andreas.abel <at> ifi.lmu.de> wrote:
You do not get them often, but here is one...

[208 of 254] Compiling Agda.TypeChecking.MetaVars.Occurs ( src/full/Agda/TypeChecking/MetaVars/Occurs.hs, dist/build/Agda/TypeChecking/MetaVars/Occurs.o )
ghc: panic! (the 'impossible' happened)
  (GHC version 7.4.1 for x86_64-unknown-linux):
        compiler/rename/RnSource.lhs:429:14-81: Irrefutable pattern failed for pattern Data.Maybe.Just (inst_tyvars,
                                                        _,
                                                        SrcLoc.L _ cls,
                                                        _)

Please report this as a GHC bug:  http://www.haskell.org/ghc/reportabug


--
Andreas Abel  <><      Du bist der geliebte Mensch.

Theoretical Computer Science, University of Munich
Oettingenstr. 67, D-80538 Munich, GERMANY

andreas.abel <at> ifi.lmu.de
http://www2.tcs.ifi.lmu.de/~abel/
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Andreas Abel | 18 Oct 19:10 2013
Picon

Re: Horray, a ghc bug!

On 18.10.2013 15:15, Stijn van Drongelen wrote:
> Have you tried the same with 7.6.1? When I search for RnSource.lhs:429
> on Google, I find multiple GHC tickets, marked as 'solved in 7.6.1'.

Thanks for the hint.  I am now compiling a ghc 7.6 (will take a while). 
Unfortunately,  7.4.1 is the last compiler with an easy binary install 
on Ubuntu 12.04 LTS, so, no quick fix.

Cheers,
Andreas

> On Fri, Oct 18, 2013 at 3:04 PM, Andreas Abel <andreas.abel <at> ifi.lmu.de
> <mailto:andreas.abel <at> ifi.lmu.de>> wrote:
>
>     You do not get them often, but here is one...
>
>     [208 of 254] Compiling Agda.TypeChecking.MetaVars.__Occurs (
>     src/full/Agda/TypeChecking/__MetaVars/Occurs.hs,
>     dist/build/Agda/TypeChecking/__MetaVars/Occurs.o )
>     ghc: panic! (the 'impossible' happened)
>        (GHC version 7.4.1 for x86_64-unknown-linux):
>              compiler/rename/RnSource.lhs:__429:14-81: Irrefutable
>     pattern failed for pattern Data.Maybe.Just (inst_tyvars,
>                                                              _,
>                                                              SrcLoc.L _ cls,
>                                                              _)
>
>     Please report this as a GHC bug:
>     http://www.haskell.org/ghc/__reportabug
>     <http://www.haskell.org/ghc/reportabug>
>
>
>     --
>     Andreas Abel  <><      Du bist der geliebte Mensch.
>
>     Theoretical Computer Science, University of Munich
>     Oettingenstr. 67, D-80538 Munich, GERMANY
>
>     andreas.abel <at> ifi.lmu.de <mailto:andreas.abel <at> ifi.lmu.de>
>     http://www2.tcs.ifi.lmu.de/~__abel/ <http://www2.tcs.ifi.lmu.de/~abel/>
>     _________________________________________________
>     Haskell-Cafe mailing list
>     Haskell-Cafe <at> haskell.org <mailto:Haskell-Cafe <at> haskell.org>
>     http://www.haskell.org/__mailman/listinfo/haskell-cafe
>     <http://www.haskell.org/mailman/listinfo/haskell-cafe>
>
>

--

-- 
Andreas Abel  <><      Du bist der geliebte Mensch.

Theoretical Computer Science, University of Munich
Oettingenstr. 67, D-80538 Munich, GERMANY

andreas.abel <at> ifi.lmu.de
http://www2.tcs.ifi.lmu.de/~abel/
Andreas Abel | 18 Oct 19:34 2013
Picon

Re: Horray, a ghc bug!

ghc-7.6.3 does produce a better error now:
[208 of 254] Compiling Agda.TypeChecking.MetaVars.Occurs ( 
src/full/Agda/TypeChecking/MetaVars/Occurs.hs, 
dist/build/Agda/TypeChecking/MetaVars/Occurs.o )

src/full/Agda/TypeChecking/MetaVars/Occurs.hs:380:10:
     Malformed instance: Occurs a -> Occurs (Elim' a)

Thanks!

On 18.10.2013 19:10, Andreas Abel wrote:
> On 18.10.2013 15:15, Stijn van Drongelen wrote:
>> Have you tried the same with 7.6.1? When I search for RnSource.lhs:429
>> on Google, I find multiple GHC tickets, marked as 'solved in 7.6.1'.
>
> Thanks for the hint.  I am now compiling a ghc 7.6 (will take a while).
> Unfortunately,  7.4.1 is the last compiler with an easy binary install
> on Ubuntu 12.04 LTS, so, no quick fix.
>
> Cheers,
> Andreas
>
>> On Fri, Oct 18, 2013 at 3:04 PM, Andreas Abel <andreas.abel <at> ifi.lmu.de
>> <mailto:andreas.abel <at> ifi.lmu.de>> wrote:
>>
>>     You do not get them often, but here is one...
>>
>>     [208 of 254] Compiling Agda.TypeChecking.MetaVars.__Occurs (
>>     src/full/Agda/TypeChecking/__MetaVars/Occurs.hs,
>>     dist/build/Agda/TypeChecking/__MetaVars/Occurs.o )
>>     ghc: panic! (the 'impossible' happened)
>>        (GHC version 7.4.1 for x86_64-unknown-linux):
>>              compiler/rename/RnSource.lhs:__429:14-81: Irrefutable
>>     pattern failed for pattern Data.Maybe.Just (inst_tyvars,
>>                                                              _,
>>                                                              SrcLoc.L
>> _ cls,
>>                                                              _)
>>
>>     Please report this as a GHC bug:
>>     http://www.haskell.org/ghc/__reportabug
>>     <http://www.haskell.org/ghc/reportabug>
>>
>>
>>     --
>>     Andreas Abel  <><      Du bist der geliebte Mensch.
>>
>>     Theoretical Computer Science, University of Munich
>>     Oettingenstr. 67, D-80538 Munich, GERMANY
>>
>>     andreas.abel <at> ifi.lmu.de <mailto:andreas.abel <at> ifi.lmu.de>
>>     http://www2.tcs.ifi.lmu.de/~__abel/
>> <http://www2.tcs.ifi.lmu.de/~abel/>
>>     _________________________________________________
>>     Haskell-Cafe mailing list
>>     Haskell-Cafe <at> haskell.org <mailto:Haskell-Cafe <at> haskell.org>
>>     http://www.haskell.org/__mailman/listinfo/haskell-cafe
>>     <http://www.haskell.org/mailman/listinfo/haskell-cafe>
>>
>>
>
>

--

-- 
Andreas Abel  <><      Du bist der geliebte Mensch.

Theoretical Computer Science, University of Munich
Oettingenstr. 67, D-80538 Munich, GERMANY

andreas.abel <at> ifi.lmu.de
http://www2.tcs.ifi.lmu.de/~abel/
Yitzchak Gale | 18 Oct 20:35 2013

Re: Horray, a ghc bug!

Andreas Abel wrote:
> ...I am now compiling a ghc 7.6 (will take a while).
> Unfortunately,  7.4.1 is the last compiler with an easy binary install on
> Ubuntu 12.04 LTS...

Hmm? We are using the binary tarball for 7.6.3 on
Ubuntu 12.04 LTS extensively. The only trouble we ever
encountered was that when installing GHC on a VM
with a very minimal install of 12.04 64-bit, we had to
temporarily install the 32-bit versions of a few basic
Ubuntu system packages to get the GHC installer to
work. After that we returned the machine to single-arch
64-bit and everything is fine.

What problems have you encountered?

-Yitz
Toby Hutton | 20 Oct 00:27 2013

Re: Horray, a ghc bug!

On Fri, Oct 18, 2013 at 02:35:02PM -0400, Yitzchak Gale wrote:
> Andreas Abel wrote:
> > ...I am now compiling a ghc 7.6 (will take a while).
> > Unfortunately,  7.4.1 is the last compiler with an easy binary install on
> > Ubuntu 12.04 LTS...
>
> Hmm? We are using the binary tarball for 7.6.3 on
> Ubuntu 12.04 LTS extensively. The only trouble we ever
> encountered was that when installing GHC on a VM
> with a very minimal install of 12.04 64-bit, we had to
> temporarily install the 32-bit versions of a few basic
> Ubuntu system packages to get the GHC installer to
> work. After that we returned the machine to single-arch
> 64-bit and everything is fine.
>
> What problems have you encountered?

I think he's saying 'apt-get install ghc' will install 7.4.1.
David Fox | 20 Oct 02:23 2013

Re: Horray, a ghc bug!

deb http://deb.seereason.com/ubuntu precise-seereason main
deb-src http://deb.seereason.com/ubuntu precise-seereason main



On Sat, Oct 19, 2013 at 3:27 PM, Toby Hutton <toby <at> grusly.com> wrote:
On Fri, Oct 18, 2013 at 02:35:02PM -0400, Yitzchak Gale wrote:
> Andreas Abel wrote:
> > ...I am now compiling a ghc 7.6 (will take a while).
> > Unfortunately,  7.4.1 is the last compiler with an easy binary install on
> > Ubuntu 12.04 LTS...
>
> Hmm? We are using the binary tarball for 7.6.3 on
> Ubuntu 12.04 LTS extensively. The only trouble we ever
> encountered was that when installing GHC on a VM
> with a very minimal install of 12.04 64-bit, we had to
> temporarily install the 32-bit versions of a few basic
> Ubuntu system packages to get the GHC installer to
> work. After that we returned the machine to single-arch
> 64-bit and everything is fine.
>
> What problems have you encountered?

I think he's saying 'apt-get install ghc' will install 7.4.1.

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Andreas Abel | 21 Oct 21:42 2013
Picon

Re: Horray, a ghc bug!

Hi, I tried the 64bit binary packages from

   haskell.org/ghc

but I got a complaint that

   libgmp.so.3

is not found on my system (and locate did not reveal it).  I have 
several of these libgmp.so, but apparently not the right one

locate libgmp.so

/usr/lib/i386-linux-gnu/openssl-1.0.0/engines/libgmp.so
/usr/lib/i386-linux-gnu/ssl/engines/libgmp.so
/usr/lib/x86_64-linux-gnu/libgmp.so
/usr/lib/x86_64-linux-gnu/libgmp.so.10
/usr/lib/x86_64-linux-gnu/libgmp.so.10.0.2
/usr/lib/x86_64-linux-gnu/openssl-1.0.0/engines/libgmp.so

Here I already gave up.  Since compiling from source worked, I did not 
investigate further...

Cheers,
Andreas

On 20.10.2013 02:23, David Fox wrote:
> deb http://deb.seereason.com/ubuntu precise-seereason main
> deb-src http://deb.seereason.com/ubuntu precise-seereason main

Ah, that is useful information.  Thanks!

> On Sat, Oct 19, 2013 at 3:27 PM, Toby Hutton <toby <at> grusly.com
> <mailto:toby <at> grusly.com>> wrote:
>
>     On Fri, Oct 18, 2013 at 02:35:02PM -0400, Yitzchak Gale wrote:
>      > Andreas Abel wrote:
>      > > ...I am now compiling a ghc 7.6 (will take a while).
>      > > Unfortunately,  7.4.1 is the last compiler with an easy binary
>     install on
>      > > Ubuntu 12.04 LTS...
>      >
>      > Hmm? We are using the binary tarball for 7.6.3 on
>      > Ubuntu 12.04 LTS extensively. The only trouble we ever
>      > encountered was that when installing GHC on a VM
>      > with a very minimal install of 12.04 64-bit, we had to
>      > temporarily install the 32-bit versions of a few basic
>      > Ubuntu system packages to get the GHC installer to
>      > work. After that we returned the machine to single-arch
>      > 64-bit and everything is fine.
>      >
>      > What problems have you encountered?
>
>     I think he's saying 'apt-get install ghc' will install 7.4.1.
>
>     _______________________________________________
>     Haskell-Cafe mailing list
>     Haskell-Cafe <at> haskell.org <mailto:Haskell-Cafe <at> haskell.org>
>     http://www.haskell.org/mailman/listinfo/haskell-cafe
>
>
>
>
> _______________________________________________
> Haskell-Cafe mailing list
> Haskell-Cafe <at> haskell.org
> http://www.haskell.org/mailman/listinfo/haskell-cafe
>

--

-- 
Andreas Abel  <><      Du bist der geliebte Mensch.

Theoretical Computer Science, University of Munich
Oettingenstr. 67, D-80538 Munich, GERMANY

andreas.abel <at> ifi.lmu.de
http://www2.tcs.ifi.lmu.de/~abel/
Jason Dagit | 21 Oct 21:58 2013
Picon

Re: Horray, a ghc bug!




On Mon, Oct 21, 2013 at 12:42 PM, Andreas Abel <andreas.abel <at> ifi.lmu.de> wrote:
Hi, I tried the 64bit binary packages from

  haskell.org/ghc

but I got a complaint that

  libgmp.so.3

is not found on my system (and locate did not reveal it).  I have several of these libgmp.so, but apparently not the right one

locate libgmp.so

It's a side effect of the distro that these packages are built on. They have a libgmp.so.3 that configure finds/uses, but I think most distros no longer install that. Often times they do have a suitable libgmp and a symlink can fix the issue.

What I do when I run into this is something roughly like this:

1. Run ldd on the executable that is complaining  (alternatively, you can use readelf -d), to find out what shared object is not being loaded correctly and get a hint of the expected path. If you can't figure out the path, then assume it will be the same path used by other .so files.

2. Find a close match, such as the ones you found with locate.

3. (Optional) install the necessary package because no close match is there.

4.  Create a symlink in the right place.

Building from source is usually easier :)

Jason
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Carter Schonwald | 21 Oct 22:13 2013
Picon

Re: Horray, a ghc bug!

indeed, isn't the usual fix to make sure you've got the "dev" version of the package name installed?


On Mon, Oct 21, 2013 at 3:58 PM, Jason Dagit <dagitj <at> gmail.com> wrote:



On Mon, Oct 21, 2013 at 12:42 PM, Andreas Abel <andreas.abel <at> ifi.lmu.de> wrote:
Hi, I tried the 64bit binary packages from

  haskell.org/ghc

but I got a complaint that

  libgmp.so.3

is not found on my system (and locate did not reveal it).  I have several of these libgmp.so, but apparently not the right one

locate libgmp.so

It's a side effect of the distro that these packages are built on. They have a libgmp.so.3 that configure finds/uses, but I think most distros no longer install that. Often times they do have a suitable libgmp and a symlink can fix the issue.

What I do when I run into this is something roughly like this:

1. Run ldd on the executable that is complaining  (alternatively, you can use readelf -d), to find out what shared object is not being loaded correctly and get a hint of the expected path. If you can't figure out the path, then assume it will be the same path used by other .so files.

2. Find a close match, such as the ones you found with locate.

3. (Optional) install the necessary package because no close match is there.

4.  Create a symlink in the right place.

Building from source is usually easier :)

Jason

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe


_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Brandon Allbery | 21 Oct 22:15 2013
Picon

Re: Horray, a ghc bug!

On Mon, Oct 21, 2013 at 4:13 PM, Carter Schonwald <carter.schonwald <at> gmail.com> wrote:
indeed, isn't the usual fix to make sure you've got the "dev" version of the package name installed?

Not here; libgmp.so.10 isn't going to match libgmp.so.3 regardless. And it's not determined while building a Haskell program, unless that program is the ghc you are going to use to build other Haskell programs.

--
brandon s allbery kf8nh                               sine nomine associates
allbery.b <at> gmail.com                                  ballbery <at> sinenomine.net
unix, openafs, kerberos, infrastructure, xmonad        http://sinenomine.net
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Jason Dagit | 21 Oct 22:16 2013
Picon

Re: Horray, a ghc bug!




On Mon, Oct 21, 2013 at 1:13 PM, Carter Schonwald <carter.schonwald <at> gmail.com> wrote:
indeed, isn't the usual fix to make sure you've got the "dev" version of the package name installed?

That works in some cases, but it's not the solution here. In this case the problem is that the elf section lists a filename that doesn't exist (libgmp.so.3) and the symlink redirects it to the real filename (libgmp.so.10).

Jason
_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Simon Hengel | 21 Oct 22:35 2013
Picon

Re: Horray, a ghc bug!

On Mon, Oct 21, 2013 at 09:42:44PM +0200, Andreas Abel wrote:
> Hi, I tried the 64bit binary packages from
> 
>   haskell.org/ghc
> 
> but I got a complaint that
> 
>   libgmp.so.3
> 
> is not found on my system (and locate did not reveal it).

Andreas, please feel free to use my PPA for GHC 7.6.3[1] and
cabal-install[2].  This assumes that you install GHC and cabal-install
over with APT and everything else with cabal-install.  Alternatively you
can also use a binary that I built on Ubuntu 13.04 [3].

Cheers.

[1] https://launchpad.net/~typeful/+archive/ghc-7.6.3
[2] https://launchpad.net/~typeful/+archive/cabal-install
[3] http://typeful.net/ghc/
Andreas Abel | 25 Oct 16:43 2013
Picon

Re: Horray, a ghc bug!

On 21.10.2013 22:35, Simon Hengel wrote:
> Andreas, please feel free to use my PPA for GHC 7.6.3[1] and
> cabal-install[2].  This assumes that you install GHC and cabal-install
> over with APT.

Simon,

I'd assume many Linux users of GHC might find this helpful.  How about 
putting this up on the GHC release pages?

I mean it is the official GHC pages where I go looking first for a 
working install, and I am expecting to find it there.  If you have a 
solution for Ubuntu (that the current pages do not provide), it would 
make sense to put it up there.

Cheers,
Andreas

--

-- 
Andreas Abel  <><      Du bist der geliebte Mensch.

Theoretical Computer Science, University of Munich
Oettingenstr. 67, D-80538 Munich, GERMANY

andreas.abel <at> ifi.lmu.de
http://www2.tcs.ifi.lmu.de/~abel/
David Fox | 22 Oct 05:28 2013

Re: Horray, a ghc bug!

The Debian libgmp package is named libgmp3 on Ubuntu, though I don't know why.  Our GHC is patched to reflect this:



On Mon, Oct 21, 2013 at 12:42 PM, Andreas Abel <andreas.abel <at> ifi.lmu.de> wrote:
Hi, I tried the 64bit binary packages from

  haskell.org/ghc

but I got a complaint that

  libgmp.so.3

is not found on my system (and locate did not reveal it).  I have several of these libgmp.so, but apparently not the right one

locate libgmp.so

/usr/lib/i386-linux-gnu/openssl-1.0.0/engines/libgmp.so
/usr/lib/i386-linux-gnu/ssl/engines/libgmp.so
/usr/lib/x86_64-linux-gnu/libgmp.so
/usr/lib/x86_64-linux-gnu/libgmp.so.10
/usr/lib/x86_64-linux-gnu/libgmp.so.10.0.2
/usr/lib/x86_64-linux-gnu/openssl-1.0.0/engines/libgmp.so

Here I already gave up.  Since compiling from source worked, I did not investigate further...

Cheers,
Andreas


On 20.10.2013 02:23, David Fox wrote:
deb http://deb.seereason.com/ubuntu precise-seereason main
deb-src http://deb.seereason.com/ubuntu precise-seereason main

Ah, that is useful information.  Thanks!

On Sat, Oct 19, 2013 at 3:27 PM, Toby Hutton <toby <at> grusly.com
<mailto:toby <at> grusly.com>> wrote:

    On Fri, Oct 18, 2013 at 02:35:02PM -0400, Yitzchak Gale wrote:
     > Andreas Abel wrote:
     > > ...I am now compiling a ghc 7.6 (will take a while).
     > > Unfortunately,  7.4.1 is the last compiler with an easy binary
    install on
     > > Ubuntu 12.04 LTS...
     >
     > Hmm? We are using the binary tarball for 7.6.3 on
     > Ubuntu 12.04 LTS extensively. The only trouble we ever
     > encountered was that when installing GHC on a VM
     > with a very minimal install of 12.04 64-bit, we had to
     > temporarily install the 32-bit versions of a few basic
     > Ubuntu system packages to get the GHC installer to
     > work. After that we returned the machine to single-arch
     > 64-bit and everything is fine.
     >
     > What problems have you encountered?

    I think he's saying 'apt-get install ghc' will install 7.4.1.

    _______________________________________________
    Haskell-Cafe mailing list
    Haskell-Cafe <at> haskell.org <mailto:Haskell-Cafe <at> haskell.org>
    http://www.haskell.org/mailman/listinfo/haskell-cafe





_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe



--
Andreas Abel  <><      Du bist der geliebte Mensch.

Theoretical Computer Science, University of Munich
Oettingenstr. 67, D-80538 Munich, GERMANY

andreas.abel <at> ifi.lmu.de
http://www2.tcs.ifi.lmu.de/~abel/

_______________________________________________
Haskell-Cafe mailing list
Haskell-Cafe <at> haskell.org
http://www.haskell.org/mailman/listinfo/haskell-cafe
Gergely Risko | 22 Oct 23:49 2013
Picon
Picon

Re: Horray, a ghc bug!

On Mon, 21 Oct 2013 21:42:44 +0200, Andreas Abel <andreas.abel <at> ifi.lmu.de> writes:

> Here I already gave up.  Since compiling from source worked, I did not
> investigate further...

If this is a single-user desktop machine and you would like to have a
modern GHC with a lot of packages built in, you can also try my pet
project called 'Ceh', see the docs here:
  - https://github.com/errge/ceh
or the slides here:
  - http://www.gergely.risko.hu/fpafternoon-zurich-20130829/ceh.html

I use this solely for all my haskell work in the last half year, even
for recompiling GHC itself, so it's pretty stable and usable by now.

Gergely

Gmane