BioHaskell / hPDB

PDB parser in Haskell
Other
20 stars 7 forks source link

Cannot compile on GHC 7.6.3 #2

Closed snoyberg closed 8 years ago

snoyberg commented 10 years ago
[ 1 of 60] Compiling Bio.PDB.Util.ParFold ( Bio/PDB/Util/ParFold.hs, dist/build/Bio/PDB/Util/ParFold.o )
[ 2 of 60] Compiling Bio.PDB.Util.MissingInstances ( Bio/PDB/Util/MissingInstances.hs, dist/build/Bio/PDB/Util/MissingInstances.o )
[ 3 of 60] Compiling Bio.PDB.EventParser.FastParse ( Bio/PDB/EventParser/FastParse.hs, dist/build/Bio/PDB/EventParser/FastParse.o )
Loading package ghc-prim ... linking ... done.
Loading package integer-gmp ... linking ... done.
Loading package base ... linking ... done.
Loading package array-0.4.0.1 ... linking ... done.
Loading package deepseq-1.3.0.1 ... linking ... done.
Loading package bytestring-0.10.0.2 ... linking ... done.
Loading package zlib-0.5.4.1 ... linking ... done.
Loading package text-1.1.1.3 ... linking ... done.
Loading package double-conversion-2.0.1.0 ... linking ... ghc: /home/ubuntu/.cabal/lib/x86_64-linux-ghc-7.6.3/double-conversion-2.0.1.0/libHSdouble-conversion-2.0.1.0.a: unknown symbol `_ZNK17double_conversion23StringToDoubleConverter12StringToIeeeIPKcEEdT_ibPi'
ghc: unable to load package `double-conversion-2.0.1.0'
Failed to install hPDB-1.2.0
cabal: Error: some packages failed to install:
hPDB-1.2.0 failed during the building phase. The exception was:
ExitFailure 1
mgajda commented 10 years ago

Hi Michael,

The library (double-conversion) has led to a trouble on a Windows platform only once in the past, did you also try the latest version? (double-conversion 2.0.6). There is -f-have-text-format flag that skips this dependency at the cost of performance.

This library compiled successfully with double-conversion and GHC 7.6.3 in the past, but it is quite possible that text-format misses some dependency constraint on version of double-conversion library at the moment.

Do you maybe use a special linker, or special compiler options that would cause "double-conversion" to fail? May I know details of your environment (env variables, platform details, etc.)

snoyberg commented 10 years ago

The only special aspects I know of in this setup are (1) I'm building into a Stackage package database, separate from my normal user database, and (2) I have the gold linker installed.

mgajda commented 10 years ago

Looks like I need to try and debug it with gold linker... Is my workaround acceptable until then? (I mean the -f-have-text-format flag.)

snoyberg commented 10 years ago

For now, I've simply disabled hPDB for GHC 7.6, so I need no other changes for now.

mgajda commented 10 years ago

As I said, hPDB worked for me on both GHC 7.6.3 and 7.8.1 (which I'm currently using.) That means that fault is almost certainly with Gold or rather ancient version of double-conversion (latest is 2.0.0.6.)

k-bx commented 10 years ago

Just wanted to note I'm having same issue, without hPDB, but with my own project relying on text-format which relies on double-conversion.

bos commented 9 years ago

This is actually due to a longstanding GHC bug that was fixed in 7.8.

mgajda commented 8 years ago

Since we have a workaround option, and the issue is fixed in 7.8, I will keep it as-is.