It’d also be interesting to compare ghc to other languages, but many languages must be installed also to run programs. E.g ocaml-base etc is installed by 80000 (if I interpret the numbers correctly), but ocaml itself, which only developers need, is only installed by 9000.

I’m not familiar with the ocaml ecosystem, though, so perhaps I misunderstand it entirely

There are two issues you are running into:

1) the interface changed between tagsoup 0.6 and 0.8
2) QuickCheck 2 is different from QC 1

In the current biolib (i.e. my darcs repo), the dependencies look like:

| Build-Depends: base>=4 && <5, QuickCheck>=2, binary==0.4.*, tagsoup>=0.4 && <0.8, bytestring >= 0.9.1,
| containers, array, parallel, parsec, random, old-time, mtl

I have also received a patch for biolib to work with both old and new tagsoup, which would obliviate the need for the upper limit to tagsoup versions.

I think the biggest problem here is that I haven’t kept up on Hackage, the 0.4 version is quite old. I’ve rolled out a new tarball (bio-0.4.4 at Hackage). You can usually get updated sources from “darcs get http://malde.org/biohaskell/biolib”

-k

you need to download and to install tagsoup 0.4

download from here:

http://hackage.haskell.org/package/tagsoup-0.4

then, you need to install using:

runhaskell Setup configure or
runhaskell Setup configure –user
runhaskell Setup build
runhaskell Setup install

so, download bio-0.4 haskell tar source package and untar package.

backup and edit bio.cabal following line:

Build-Depends: base>=3 && =1.2.0.0, binary, tagsoup= 0.9.1, containers, array,
parallel, parsec, random, old-time, mtl

I changed QuickCheck to >=1.2 and tagsoup =1.2.0.0: using QuickCheck-2.1.0.3
Dependency tagsoup <=0.4: using tagsoup-0.4

ok, good then install biohaskell lib using:

runhaskell Setup configure –user
runhaskell Setup build

opps so, you will see and error

[18 of 43] Compiling Bio.Sequence.TwoBit ( Bio/Sequence/TwoBit.hs, dist/build/Bio/Sequence/TwoBit.o )

Bio/Sequence/TwoBit.hs:37:31:
Module `Test.QuickCheck' does not export `check'

to fix error open Bio/Sequence/TwoBit.hs haskell source file and locate line 37 with your vi or emacs editor

by default you will see:

import Test.QuickCheck hiding (check) — QC 1.0
–import Test.QuickCheck hiding ((.&.)) — QC 2.0

please make changes like this:

– import Test.QuickCheck hiding (check) — QC 1.0
import Test.QuickCheck hiding ((.&.)) — QC 2.0

this is because I am using QuickCheck 2

ok, good , try to build once again :

runhaskell Setup build

opps error

[35 of 43] Compiling Bio.Util.TestBase ( Bio/Util/TestBase.hs, dist/build/Bio/Util/TestBase.o )

Bio/Util/TestBase.hs:81:4:
`coarbitrary' is not a (visible) method of class `Arbitrary'

Bio/Util/TestBase.hs:85:4:
`coarbitrary' is not a (visible) method of class `Arbitrary'

Bio/Util/TestBase.hs:90:4:
`coarbitrary' is not a (visible) method of class `Arbitrary'

Bio/Util/TestBase.hs:98:4:
`coarbitrary' is not a (visible) method of class `Arbitrary'

Bio/Util/TestBase.hs:105:4:
`coarbitrary' is not a (visible) method of class `Arbitrary'

Bio/Util/TestBase.hs:109:4:
`coarbitrary' is not a (visible) method of class `Arbitrary'

Bio/Util/TestBase.hs:117:4:
`coarbitrary' is not a (visible) method of class `Arbitrary'

Bio/Util/TestBase.hs:125:4:
`coarbitrary' is not a (visible) method of class `Arbitrary'

Bio/Util/TestBase.hs:132:4:
`coarbitrary' is not a (visible) method of class `Arbitrary'

you will need to edit bio.cabal, find Bio.Util.TestBase and delete it then save file.

runhaskell Setup install
Installing library in $HOME/.cabal/lib/bio-0.4/ghc-6.12.1
Registering bio-0.4…

ls $HOME/.cabal/lib/bio-0.4/ghc-6.12.1/
Bio HSbio-0.4.o libHSbio-0.4.a

ready, that's all

ketil, what's is your opinion ?

Ciao

Configuring bio-0.4.4...
Dependency QuickCheck >=2: using QuickCheck-2.1.0.2
Dependency array -any: using array-0.3.0.0
Dependency base ==4.*: using base-4.2.0.0
Dependency binary -any: using binary-0.5.0.2
Dependency bytestring >=0.9.1: using bytestring-0.9.1.5
Dependency containers -any: using containers-0.3.0.0
Dependency mtl -any: using mtl-1.1.0.2
Dependency old-time -any: using old-time-1.0.0.3
Dependency parallel -any: using parallel-1.1.0.1
Dependency parsec -any: using parsec-2.1.0.1
Dependency random -any: using random-1.0.0.2
Dependency tagsoup >=0.4: using tagsoup-0.6


Bio/Alignment/BlastXML.hs:41:12:
`Tag’ is not applied to enough type arguments
Expected kind `*’, but `Tag’ has kind `* -> *’
In the type signature for `getFrom’:
getFrom :: [Tag] -> String -> String

Could you please perhaps state clearly which combination of the complier/libraries is recommended?

cabal install bio
Resolving dependencies…
cabal: cannot configure tagsoup-0.8. It requires QuickCheck ==2.1.*
For the dependency on QuickCheck ==2.1.* there are these packages:
QuickCheck-2.1, QuickCheck-2.1.0.1, QuickCheck-2.1.0.2 and QuickCheck-2.1.0.3.
However none of them are available.
QuickCheck-2.1 was excluded because bio-0.4 requires QuickCheck <2

Is there anything to be done to fix this?
j131

I did a quick test now, and on my laptop (1.3GHz Core2), it took exactly one minute to process 30K ESTs (about 23MB of sequence, and 60MB quality data). Memory use is constant at about 20MB, according to ‘top’ (using lazy bytestrings for the sequence and qual data -> nice streaming behavior)

So: about 500 ESTs/second at the moment. With some tuning, and perhaps using multiple threads, it should be possible to make parsing integers in the quality data the bottleneck. But this is fast enough for me.