Bug#889623: kraken: failing autopkgtest, possibly broken package

2018-02-07 Thread Andreas Tille
Hi again,

the issue is most probably caused by a broken fasta parser.  With the
following simplification of the header


 $ git diff
diff --git a/debian/tests/test_data/Acartia_tonsa.fasta 
b/debian/tests/test_data/Acartia_tonsa.fasta
index abcda65..011076b 100644
--- a/debian/tests/test_data/Acartia_tonsa.fasta
+++ b/debian/tests/test_data/Acartia_tonsa.fasta
@@ -1,4 +1,4 @@
->gi|441431932| Acartia tonsa copepod circovirus isolate 154_D11, complete 
genome
+>441431932
 ACCTCGGCAACATCCGATCATCATATGATCAATTATGACTCATCCCGCGTGGAATTATGTAGCCAATGAA
 ATCGCTCCATATTTCATTGAGCACAGTGGCCGCAATCTATAAAGCCGCGAGCGAAGCGAGCG
 GTAGGCACAGTTTGACCTGCCTAGCAACGCAACAACCAACCGAGCCCGTGGGTGGTGCTTCA
diff --git a/debian/tests/test_data/Acinetobacter_phage.fasta 
b/debian/tests/test_data/Acinetobacter_phage.fasta
index 317a94c..591f2e0 100644
--- a/debian/tests/test_data/Acinetobacter_phage.fasta
+++ b/debian/tests/test_data/Acinetobacter_phage.fasta
@@ -1,4 +1,4 @@
->gi|28173057| Acinetobacter phage AP205, complete genome
+>28173057
 GGAGTGAAGGAGTTCGCTGAAAGCCGAATCGAATTCGACTTTGCGTGATTCACATCACGTCT
 TACTCACGATACTAGTACCGCGAGTTATCTTGTGGTAATTACTACCAGGAGATAACTTTATGAAGA
 AAAGGACGCCTTGCTTCCCTATGCGGCATCATACTCAGCTTTCAACTAACATTGTTGACTGC



the test works again as expected


$ sh /usr/share/doc/kraken/run-unit-test
Added "Acartia_tonsa.fasta" to library (test_db)
Added "Acinetobacter_phage.fasta" to library (test_db)
Kraken build set to minimize disk writes.
Creating k-mer set (step 1 of 6)...
Found jellyfish v1.1.11
Hash size not specified, using '5120'
K-mer set created. [0.029s]
Skipping step 2, no database reduction requested.
Sorting k-mer set (step 3 of 6)...
K-mer set sorted. [0.030s]
Skipping step 4, GI number to seqID map now obsolete.
Creating seqID to taxID map (step 5 of 6)...
3 sequences mapped to taxa. [0.009s]
Setting LCAs in database (step 6 of 6)...
Finished processing 3 sequences
Database LCAs set. [0.020s]
Database construction complete. [Total: 0.105s]
2 sequences (0.00 Mbp) processed in 0.000s (983.6 Kseq/m, 94.43 Mbp/m).
  2 sequences classified (100.00%)
  0 sequences unclassified (0.00%)
  0.00  0   0   U   0   unclassified


Since we want to make sure our users will be able to work with usual
fasta files I'll open an issue upstream and leave the bug open.

Kind regards

   Andreas. 

-- 
http://fam-tille.de



Bug#889623: kraken: failing autopkgtest, possibly broken package

2018-02-07 Thread Andreas Tille
Hi,

since upstream claimed to have fixed a bug unrelated to this test case I
meanwhile uploaded kraken 1.1.  To simplify fixing the issue here is the
output of the autopkgtest script on a local machine:


$ sh /usr/share/doc/kraken/run-unit-test
scan_fasta_file.pl: unable to determine taxonomy ID for sequence gi|441431932|
scan_fasta_file.pl: unable to determine taxonomy ID for sequence gi|28173057|
Kraken build set to minimize disk writes.
Creating k-mer set (step 1 of 6)...
Found jellyfish v1.1.11
kmer_estimator: malformed fasta file - expected header char > not found
Hash size not specified, using '2560'
terminate called after throwing an instance of 
'jellyfish::file_parser::FileParserError'
  what():  Empty input file '/dev/fd/0': Success
/usr/lib/kraken/build_kraken_db.sh: Zeile 96:  4909 Fertig  
find library/ -name '*.fna' -print0
  4910   | xargs -0 cat
  4911 Abgebrochen | jellyfish1 count -m $KRAKEN_KMER_LEN -s 
$KRAKEN_HASH_SIZE -C -t $KRAKEN_THREAD_CT -o database /dev/fd/0
kraken: database ("./test_db") does not contain necessary file database.kdb
kraken-report: database ("./test_db") does not contain necessary file 
database.kdb


It would be really great if somebody could have a look

  Andreas.


-- 
http://fam-tille.de



Bug#889623: Please fix autopkgtest (Was: Bug#889623: kraken: failing autopkgtest, possibly broken package)

2018-02-05 Thread Andreas Tille
Hi,

On Sun, Feb 04, 2018 at 08:13:48PM -0800, Steve Langasek wrote:
> Since the upload of kraken 1.0-1, the autopkgtests for this package
> have been consistently failing in both Debian and Ubuntu.

I confirm that the test is failing.  I commited version 1.1 of kraken to
Git and it keeps on failing.  I assume that upstream might have changed
the interface. Nadiya (or anybody else who has some background with
biological applications) could you please have a look?

Kind regards

   Andreas.

-- 
http://fam-tille.de



Bug#889623: kraken: failing autopkgtest, possibly broken package

2018-02-04 Thread Steve Langasek
Source: kraken
Version: 1.0-1
Severity: important
User: ubuntu-de...@lists.ubuntu.com
Usertags: origin-ubuntu bionic autopkgtest

Dear Andreas,

Since the upload of kraken 1.0-1, the autopkgtests for this package
have been consistently failing in both Debian and Ubuntu.  While the test
failure is (unfortunately) not considered a blocker for Debian testing,
autopkgtest regressions are blockers for Ubuntu releases.

autopkgtest [18:36:18]: test run-unit-test: [---
scan_fasta_file.pl: unable to determine taxonomy ID for sequence gi|441431932|
autopkgtest [18:36:19]: test run-unit-test: ---]
autopkgtest [18:36:19]: test run-unit-test:  - - - - - - - - - - results - - - 
- - - - - - -
run-unit-testFAIL non-zero exit status 25

  https://ci.debian.net/packages/k/kraken/unstable/amd64/

There is very little output from the test, so I'm unsure whether this
represents serious breakage in the package; but I do see that the tests
themselves have not changed from the previous version of the package where
the test did pass, so it does seem likely to me that this is a bug in the
package rather than in the test.

-- 
Steve Langasek   Give me a lever long enough and a Free OS
Debian Developer   to set it on, and I can move the world.
Ubuntu Developerhttp://www.debian.org/
slanga...@ubuntu.com vor...@debian.org


signature.asc
Description: PGP signature