Failed test 'header_text'

[boxarocks]

C:\>perl -v

This is perl 5, version 20, subversion 1 (v5.20.1) built for MSWin32-x64-multi-thread


t/026FileApp.t ............ 1/27
#   Failed test 'header_text'
#   at t/026FileApp.t line 459.
#          got: 'This is a nice header.
# This is a nice header.
# DEBUG - waah!
# '
#     expected: 'This is a nice header.
# DEBUG - waah!
# '
# Looks like you failed 1 test of 27.
t/026FileApp.t ............ Dubious, test returned 1 (wstat 256, 0x100)
Failed 1/27 subtests
        (less 8 skipped subtests: 18 okay)



Test Summary Report
-------------------
t/026FileApp.t          (Wstat: 256 Tests: 27 Failed: 1)
  Failed test:  23
  Non-zero exit status: 1
Files=73, Tests=740, 35 wallclock secs ( 0.44 usr +  0.05 sys =  0.48 CPU)
Result: FAIL
Failed 1/73 test programs. 1/740 subtests failed.
dmake.exe:  Error code 255, while making 'test_dynamic'
  MSCHILLI/Log-Log4perl-1.48.tar.gz
  C:\STRAWB~1\c\bin\dmake.exe test -- NOT OK
//hint// to see the cpan-testers results for installing this module, try:
  reports MSCHILLI/Log-Log4perl-1.48.tar.gz
Stopping: 'install' failed for 'Log::Log4perl'.

[iafan]

Same here — doesn't install on Windows for the same reason. My Perl version:

perl -v

This is perl 5, version 16, subversion 1 (v5.16.1) built for MSWin32-x64-multi-thread

[prat0088]

I can also reproduced the failure using Log::Log4perl 1.48.

perl -v

This is perl 5, version 24, subversion 0 (v5.24.0) built for MSWin32-x64-multi-thread

It looks like #73 is related.

I'm looking at log(), but I don't see any reason in the code why it would print the header twice only on Windows using syswrite(). I'm guessing there may be a subtle difference in the behavior of a system call between Windows and Linux, but I don't have time to dig into it right now.

[mschilli]

#73 is fixed, please re-test and re-open if it's still an issue on Windows (which I unfortunately don't have easy access to).

[briandfoy]

I'm getting this issue with 1.49 on AppVeyor (a free CI system for Windows for those of us without our own Windows boxes):

https://ci.appveyor.com/project/briandfoy/ghojo/build/1.0.21

[itsuki-hayashi]

@mschilli The issue does not relate to the use of syswrite, it is just that the unlink on Windows works differently than Unix, it might be an issue with the Perl CORE.

If we use overwrite instead of append when opening the test file, the test will success.
This pull request fixed the issue:
#100