testsuite failure with latest Debian and yaml packages.

[LocutusOfBorg]

Hello, following the test failure, occurring also with 0.8.23.1
https://buildd.debian.org/status/package.php?p=haskell-yaml&suite=unstable

do you have any idea?

. /usr/share/haskell-devscripts/Dh_Haskell.sh && \
check_recipe
Running debian/hlibrary.setup test --builddir=dist-ghc --show-details=direct
Running 1 test suites...
Test suite spec: RUNNING...

Data.Yaml.Include
  decodeFile
    supports includes
    supports recursive includes
    aborts on cyclic includes
    when file does not exist
      throws Left (InvalidYaml (Just (YamlException "Yaml file not found: ...")))
  decodeFileEither
    when file does not exist
      returns Left (InvalidYaml (Just (YamlException "Yaml file not found: ...")))
Data.Yaml
  streaming
    count scalars with anchor
    count sequences with anchor
    count mappings with anchor
    count aliases
    count scalars
    largest string
    encode/decode
    encode/decode file
    interleaved encode/decode
    decode invalid document (without segfault)
  Data.Yaml
    encode/decode
    encode/decode file
    encode/decode files with non-ASCII names
    encode/decode strings
    decode invalid file
    processes datatypes
    encode invalid numbers FAILED [1]
  Data.Yaml.Pretty
    encode/decode
    encode/decode strings
    processes datatypes
  Data.Yaml aliases
    simple scalar alias
    simple sequence alias
    simple mapping alias
    mapping alias before anchor
    mapping alias inside anchor
    scalar alias overriding
  Data.Yaml merge keys
    test uniqueness of keys
    test mapping merge
    test sequence of mappings merging
  numbers
    parses as string when quoted
    parses as number when unquoted
    parses as number !!str is present
    integers have no decimals FAILED [2]
  booleans
    parses when all lowercase
    parses when all uppercase
    parses when titlecase
  empty input
    parses as Null
    parses as Null from file
  comment only input
    parses as Null
    parses as Null from file
  alias stripping
    works
  nulls
    null recognized: "null"
    null recognized: "Null"
    null recognized: "NULL"
    null recognized: "~"
    null recognized: ""
  pretty output
    simple nulls
    simple numbers
    True
    False
    simple string
  special keys
    fo"o
    fo'o
    fo\'o
    fo: o
    foo
bar
baz

  special values
    fo"o
    fo'o
    fo\'o
    fo: o
    foo
bar
baz

  decodeFileEither
    loads YAML through JSON into Haskell data
    when file does not exist
      returns Left
  round-tripping of special scalars
    y
    Y
    On
    ON
    false
    12345
    12345.0
    12345a
    12e3
    no tags
  aliases in keys #49
  serialization of +123 #64
  preserves Scientific precision
  truncates files

Failures:

  test/Data/YamlSpec.hs:440: 
  1) Data.Yaml.Data.Yaml encode invalid numbers
       expected: ".\n"
        but got: ".\n...\n"

  test/Data/YamlSpec.hs:458: 
  2) Data.Yaml.numbers integers have no decimals
       expected: "1\n"
        but got: "1\n...\n"

Randomized with seed 1593000096

Finished in 0.0111 seconds
82 examples, 2 failures
Test suite spec: FAIL
Test suite logged to: dist-ghc/test/yaml-0.8.23.1-spec.log
0 of 1 test suites (0 of 1 test cases) passed.
/usr/share/cdbs/1/class/hlibrary.mk:154: recipe for target 'check-ghc-stamp' failed
make: *** [check-ghc-stamp] Error 1

[snoyberg]

I'm guessing because we have a more recent (from Git) version of the library locally embedded, and Debian is still using the officially released, unpatched version of the library. The libyaml situation is pretty murky to me from what I've seen of the discussion.

…

On Fri, Jun 23, 2017 at 5:02 PM, Gianfranco Costamagna < ***@***.***> wrote: Hello, following the test failure, occurring also with 0.8.23.1 https://buildd.debian.org/status/package.php?p=haskell-yaml&suite=unstable do you have any idea? . /usr/share/haskell-devscripts/Dh_Haskell.sh && \ check_recipe Running debian/hlibrary.setup test --builddir=dist-ghc --show-details=direct Running 1 test suites... Test suite spec: RUNNING... Data.Yaml.Include decodeFile supports includes supports recursive includes aborts on cyclic includes when file does not exist throws Left (InvalidYaml (Just (YamlException "Yaml file not found: ..."))) decodeFileEither when file does not exist returns Left (InvalidYaml (Just (YamlException "Yaml file not found: ..."))) Data.Yaml streaming count scalars with anchor count sequences with anchor count mappings with anchor count aliases count scalars largest string encode/decode encode/decode file interleaved encode/decode decode invalid document (without segfault) Data.Yaml encode/decode encode/decode file encode/decode files with non-ASCII names encode/decode strings decode invalid file processes datatypes encode invalid numbers FAILED [1] Data.Yaml.Pretty encode/decode encode/decode strings processes datatypes Data.Yaml aliases simple scalar alias simple sequence alias simple mapping alias mapping alias before anchor mapping alias inside anchor scalar alias overriding Data.Yaml merge keys test uniqueness of keys test mapping merge test sequence of mappings merging numbers parses as string when quoted parses as number when unquoted parses as number !!str is present integers have no decimals FAILED [2] booleans parses when all lowercase parses when all uppercase parses when titlecase empty input parses as Null parses as Null from file comment only input parses as Null parses as Null from file alias stripping works nulls null recognized: "null" null recognized: "Null" null recognized: "NULL" null recognized: "~" null recognized: "" pretty output simple nulls simple numbers True False simple string special keys fo"o fo'o fo\'o fo: o foo bar baz special values fo"o fo'o fo\'o fo: o foo bar baz decodeFileEither loads YAML through JSON into Haskell data when file does not exist returns Left round-tripping of special scalars y Y On ON false 12345 12345.0 12345a 12e3 no tags aliases in keys #49 serialization of +123 #64 preserves Scientific precision truncates files Failures: test/Data/YamlSpec.hs:440: 1) Data.Yaml.Data.Yaml encode invalid numbers expected: ".\n" but got: ".\n...\n" test/Data/YamlSpec.hs:458: 2) Data.Yaml.numbers integers have no decimals expected: "1\n" but got: "1\n...\n" Randomized with seed 1593000096 Finished in 0.0111 seconds 82 examples, 2 failures Test suite spec: FAIL Test suite logged to: dist-ghc/test/yaml-0.8.23.1-spec.log 0 of 1 test suites (0 of 1 test cases) passed. /usr/share/cdbs/1/class/hlibrary.mk:154: recipe for target 'check-ghc-stamp' failed make: *** [check-ghc-stamp] Error 1 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#110>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AADBB7nVPl0Jkv1KPI869GYoDkvrL-phks5sG8V6gaJpZM4ODm0i> .

[LocutusOfBorg]

I'm not sure I follow the library you are referring to:
https://bitbucket.org/xi/libyaml/commits/branch/default
this seems the same version/commit as our libyaml-dev library
https://packages.qa.debian.org/liby/libyaml.html

do you mean something else?

[snoyberg]

Then I'm not sure what's wrong. I'd recommend trying to run a build with the non-system libyaml to see if it resolves the test failures.

…

On Fri, Jun 23, 2017 at 5:31 PM, Gianfranco Costamagna < ***@***.***> wrote: I'm not sure I follow the library you are referring to: https://bitbucket.org/xi/libyaml/commits/branch/default this seems the same version/commit as our libyaml-dev library https://packages.qa.debian.org/liby/libyaml.html do you mean something else? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#110 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AADBBwRdfq7M4Zgy2hMalxzkV5WmeDCqks5sG8xNgaJpZM4ODm0i> .

[LocutusOfBorg]

--- libyaml-0.1.7.orig/src/emitter.c
+++ libyaml-0.1.7/src/emitter.c
@@ -1946,10 +1946,6 @@ yaml_emitter_write_plain_scalar(yaml_emi

     emitter->whitespace = 0;
     emitter->indention = 0;
-    if (emitter->root_context)
-    {
-        emitter->open_ended = 1;
-    }

     return 1;
 }

I did a diff between your embedded yaml and the debian one and traced down the differences, until I found the culprit, the above patch, applied to yaml, makes it pass the testsuite
thanks!

[LocutusOfBorg]

sure, the development moved to github, and the commit is here
yaml/libyaml@56400d9

thanks!

[LocutusOfBorg]

Can you please consider applying the latest patch posted on the issue I opened upstream?
yaml/libyaml#60
It seems to be the correct approach to fix the testsuite

[snoyberg]

I've pushed a commit to do the update, are you able to test it before I make a Hackage release?

[LocutusOfBorg]

actually the patch is not mainline yet, so your sync is mostly useless.
the mainline upstream libyaml has still the bad commit, that just by chance, makes the testsuite pass

[LocutusOfBorg]

Hello, the new libyaml release is out, can you please sync libyaml again? :)

[snoyberg]

Reopening an old issue that's (seemingly) unrelated to the current request isn't good practice. In the future, could you open a new issue?

I can sync against the upstream repo. From what I can see, the newest commit is f6e09f829b606ca0d0adf774236fa8cdd5e2a7d1, and is the 0.2.1 release.

[LocutusOfBorg]

thanks, and yes, I'll open a new one in case :)