[Pkg-haskell-maintainers] Bug#725004: haskell-cryptocipher: FTBFS on mips, ppc, s390 and sparc (but was built there in the past)

[Niels Thykier]

Control: tags -1 confirmed
Control: retitle -1 haskell-cryptocipher: FTBFS on big-endian archs

On 2013-10-02 07:27, Niels Thykier wrote:
> On 2013-09-30 14:12, Niels Thykier wrote:
>> Package: haskell-cryptocipher
>> Version: 0.5.2-1
>> Severity: serious
>>
>> Hi,
>>
>> The buildd logs all appear to have a log that looks something like:
>>
>> """
>> *** Failed! Falsifiable (after 1 test): 
>> DESEDE2Message "~\134\162\184\232O\193\150n{#\180m\SYNi\240" "3;\157\243<Q\a\SYN\SOH6[\138D\129Og\134\143m\rWE8\188\157k7Y\171A\n\DC1" "j\248\217e\166O|c"
>> (used seed 9713386)
>>   DES-EDE2 (CTR): [OK, passed 100 tests]
>>   DES-EDE2 (CFB): [OK, passed 100 tests]
>>   DES-EDE2 (OFB): [OK, passed 100 tests]
>>
>>          Properties   Total       
>>  Passed  15           15          
>>  Failed  8            8           
>>  Total   23           23          
>> Test suite test-cryptocipher: FAIL
>> Test suite logged to: dist-ghc/test/cryptocipher-0.5.2-test-cryptocipher.log
>> 0 of 1 test suites (0 of 1 test cases) passed.
>> make: *** [check-ghc-stamp] Error 1
>> """
>>
>> These build failures are currently preventing haskell-cryptocipher
>> from migrating to testing (and quite possibly blocking several other
>> haskell packages as well).
>>
>> ~Niels
>>
> 
> Hey,
> 
> On IRC, Joey Hess suspected this failure might be related to endianess
> based on the failing architectures appearing to be big endian (and that
> "cryptocipher certainly does some low-level bit-banging in C").
> 
> ~Niels
> 
> 

Tagging confirmed based on Joey's review:

"""
06:29 < joeyh> ok, here's the bug
06:29 < joeyh> word64ToBs :: Word64 -> B.ByteString
06:29 < joeyh> word64ToBs = runPut . putWord64le
06:29 < joeyh> this has all been extensively rewritten in newer upstream
releases, and I think, fixed for big endian
"""

~Niels