Fixes for big-endian platforms

[mosra]

Should cover everything reported in #58 (and https://pastebin.com/GkSgVGyd), except the System test, as I have no idea what preprocessor macros to use to detect a Power9 architecture. I'm also a bit unsure about the SHA-1 fix, but hopefully it's the correct place to make that change.

@pkubaj would be wonderful if you could run the tests again on this branch 🙏 thanks in advance!

[codecov-io]

Codecov Report

Merging #60 into master will not change coverage.
The diff coverage is 100%.

@@           Coverage Diff           @@
##           master      #60   +/-   ##
=======================================
  Coverage   98.02%   98.02%           
=======================================
  Files          79       79           
  Lines        5105     5105           
=======================================
  Hits         5004     5004           
  Misses        101      101

Impacted Files	Coverage Δ
src/Corrade/Utility/Sha1.cpp	100% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 12a0fd7...e20d61a. Read the comment docs.

[pkubaj]

I'm testing it right now (UtilityDirectoryTest has been running for the last 10 minutes, is that normal)?

One test that failed so far is:

15/67 Test #15: ContainersStridedArrayViewTest .........***Failed    0.12 sec
Starting Corrade::Containers::Test::StridedArrayViewTest with 22 test cases...
    OK [01] constructEmpty()
    OK [02] constructNullptr()
    OK [03] constructNullptrSize()
    OK [04] construct()
    OK [05] constructFixedSize()
    OK [06] constructDerived()
    OK [07] constructView()
    OK [08] constructStaticView()
    OK [09] convertBool()
    OK [10] convertConst()
    OK [11] convertFromExternalView()
    OK [12] convertConstFromExternalView()
    OK [13] emptyCheck()
    OK [14] access()
    OK [15] accessConst()
    OK [16] accessInvalid()
    OK [17] iterator()
    OK [18] rangeBasedFor()
    OK [19] sliceInvalid()
    OK [20] slice()
  FAIL [21] cast() at /usr/local/poudriere/ports/default/devel/corrade/work/corrade-2019.01/src/Corrade/Containers/Test/StridedArrayViewTest.cpp on line 589
        Values b[2] and (4 << 16) | 40 are not the same, actual is
        196638
        but expected
        262184
    OK [22] castInvalid()
Finished Corrade::Containers::Test::StridedArrayViewTest with 1 errors out of 173 checks.

You won't able to test for POWER9 (this is like testing specifically for Skylake or Ryzen, you need to ping kernel directly for that), but you can test for POWER:

root@talos:$~$ echo | g++8 -dM -E - | egrep -i 'ppc|powerpc'
#define _ARCH_PPCGR 1
#define __PPC64__ 1
#define _ARCH_PPCSQ 1
#define _ARCH_PPC 1
#define __ppc__ 1
#define __powerpc64__ 1
#define __PPC__ 1
#define __powerpc__ 1
#define _ARCH_PPC64 1

[pkubaj]

OK, it finished.

The only failing tests are:

15/67 Test #15: ContainersStridedArrayViewTest .........***Failed    0.12 sec
Starting Corrade::Containers::Test::StridedArrayViewTest with 22 test cases...
    OK [01] constructEmpty()
    OK [02] constructNullptr()
    OK [03] constructNullptrSize()
    OK [04] construct()
    OK [05] constructFixedSize()
    OK [06] constructDerived()
    OK [07] constructView()
    OK [08] constructStaticView()
    OK [09] convertBool()
    OK [10] convertConst()
    OK [11] convertFromExternalView()
    OK [12] convertConstFromExternalView()
    OK [13] emptyCheck()
    OK [14] access()
    OK [15] accessConst()
    OK [16] accessInvalid()
    OK [17] iterator()
    OK [18] rangeBasedFor()
    OK [19] sliceInvalid()
    OK [20] slice()
  FAIL [21] cast() at /usr/local/poudriere/ports/default/devel/corrade/work/corrade-2019.01/src/Corrade/Containers/Test/StridedArrayViewTest.cpp on line 589
        Values b[2] and (4 << 16) | 40 are not the same, actual is
        196638
        but expected
        262184
    OK [22] castInvalid()
Finished Corrade::Containers::Test::StridedArrayViewTest with 1 errors out of 173 checks.

55/67 Test #55: UtilitySystemTest ......................***Failed    0.10 sec
Starting Corrade::Utility::Test::SystemTest with 2 test cases...
  FAIL [1] target() at /usr/local/poudriere/ports/default/devel/corrade/work/corrade-2019.01/src/Corrade/Utility/Test/SystemTest.cpp on line 45
        Expression !"No suitable CORRADE_TARGET_* defined on this platform" failed.
    OK [2] sleep()
Finished Corrade::Utility::Test::SystemTest with 1 errors out of 2 checks.

The 1st is mentioned above, the 2nd you can fix by checking e.g. for __powerpc64__.

[mosra]

Thanks a lot! 👍 The StridedArrayViewTest was just a silly copypaste typo, the System test should be working now as well, I added a CORRADE_TARGET_POWERPC macro that checks for __powerpc64__ (or _M_PPC for MSVC, but I have no idea if that's still a thing).

UtilityDirectoryTest has been running for the last 10 minutes, is that normal?

Yeah sorry there's a little benchmark comparing various file copy approaches. It runs in a second on a SSD and in around five seconds on a HDD. Didn't expect it to run on such a device, so that's why 😅

I force-pushed an updated version, if you could do one last check with just the two formerly failing tests, like this, would be great. Thanks for all your help! 👍

ctest --output-on-failure -R (SystemTest|StridedArrayViewTest)

[pkubaj]

All tests have passed.
Thanks!

[mosra]

Likewise, thanks for the confirmation! :)