Bug#863015: [Debian-med-packaging] Bug#863015: Patch for reproducible builds breaks build time test

2017-08-27 Thread Chris Lamb 
Hi Andreas,

  7 --- a/include/jellyfish/generic_file_header.hpp
  8 +++ b/include/jellyfish/generic_file_header.hpp
  9 @@ -180,6 +180,8 @@ public:
 10 
 11  protected:
 12std::string get_hostname() const {
 13 +if(std::getenv("SOURCE_DATE_EPOCH"))
 14 +  return "";
  ^^

Just change this to "hostname".

The test was checking whether it was not the empty string. Thinking about
it, returning an empty string is probably not a good idea anyway.

Enjoy :)


Best wishes,

-- 
  ,''`.
 : :'  : Chris Lamb
 `. `'`  la...@debian.org / chris-lamb.co.uk
   `-

Bug#863015: [Debian-med-packaging] Bug#863015: Patch for reproducible builds breaks build time test

2017-08-27 Thread Chris Lamb 
Hi Andreas,

> # FAIL:  1

Just pasting this here for now so I don't lose it. Here is the entry
in unit_tests/unit_tests.log after a failed build:

   543 [--] 2 tests from FileHeader
   544 [ RUN  ] FileHeader.Standard
   545 unit_tests/test_file_header.cc:26: Failure
   546 Expected: ("") != (h["hostname"]), actual: "" vs ""
   547 [  FAILED  ] FileHeader.Standard (0 ms)
   548 [ RUN  ] FileHeader.WriteRead
   549 [   OK ] FileHeader.WriteRead (1 ms)
   550 [--] 2 tests from FileHeader (1 ms total)

(More later unless you fix this first...)


Regards,

-- 
  ,''`.
 : :'  : Chris Lamb
 `. `'`  la...@debian.org / chris-lamb.co.uk
   `-

Bug#863015: Patch for reproducible builds breaks build time test

2017-08-25 Thread Andreas Tille 
Hi Chris,

thanks for the patch (and sorry for taking so much time to respond).
Unfurtunately if I activate the patch[1] one test
(unit_tests/unit_tests.sh) fails which is not the case if the patch
is deactivated in debian/patches/series:

...
PASS: tests/large_key.sh
PASS: tests/parallel_hashing.sh
FAIL: unit_tests/unit_tests.sh
===
   jellyfish 2.2.6: ./test-suite.log
===

# TOTAL: 13
# PASS:  9
# SKIP:  3
# XFAIL: 0
# FAIL:  1
# XPASS: 0
# ERROR: 0
...


It might be that the test relies somehow on the date in the json file.

Should I report this to upstream or do you want to investigate into the
test suite?  I admit I'm overwhelmed with gcc-7 and other bugs and will
not manage to track this down in any sensible time frame.

Kind regards and thanks in any case for the reprodicible build effort

 Andreas.


[1] 
https://anonscm.debian.org/git/debian-med/jellyfish.git/tree/debian/patches/reproducible.patch

-- 
http://fam-tille.de