"internal error" in 6-2-branch

[Serge D. Mechveliani]

Dear GHC team,

I keep on trying to build a correct  ghc-6-2-branch  
from cvs source.
This is a pain. Spent a week to this, and no success.
Just
     "Internal error ..."

The last experience was on computer under  Debian linux,
with 
     no ghc, no alex, no happy  anywhere. 

I installed there  ghc-6.2.1 binary  to
                                  /home/mechvel/ghc/6.2.1/inst

Then, got  cvs  ghc-6-2-branch  to 
                                /home/mechvel/ghc/cvs/6-2-branch/fptools
,
installed  alex,  happy  separately, with  ghc-6.2.1.
,
`made'  6-2-branch with  ghc-6.2.1,  installed it to
                                /home/mechvel/ghc/cvs/6-2-branch/inst/

(this time it compiled without adventures).
,

erased  ghc-6.2.1   !  -- no ghc, no ghci  remain, except of
                                     /home/mechvel/ghc/cvs/6-2-branch/

,
tested:  > which ghc
           /home/mechvel/ghc/cvs/6-2-branch/inst/bin/ghc
         > which ghci
           /home/mechvel/ghc/cvs/6-2-branch/inst/bin/ghci
         >
,
`made' DoCon by the Makefile containing 
                    ghc = /home/mechvel/ghc/cvs/6-2-branch/inst/bin/ghc
                    ...

`made' the test  T_  by     > ghc $pcpdocon --make T_ 
,

run the test by             > ghci $pcpdocon T_ 
and got
        "Internal error ..."

-- a message I keep on reporting to you, appearing somewhere in 
the middle of the test.

And it looks like you cannot reproduce the effect.

Maybe, I'll try to remake it with 6-2-branch.
But the whole story looks strange, does not it?

With kind regards,

-----------------
Serge Mechveliani
mechvel at botik.ru