[1.1.9] sas7bcat regression on big-endian architectures

[pinotree]

test_readstat in ReadStat 1.1.9 fails on big-endian architectures for sas7bcat; I noticed this issue when uploading 1.1.9 to Debian unstable (see the build logs of the various big-endian architectures: [1][2][3][4][5]). The failed test log is:

FAIL: test_readstat
===================

Writing file buffer to /tmp/test_readstat.sas7bcat
Test "" failed: Value labels count
 * Format: sas7bcat (0x100000)
 * Expected: 2.000000 (double)
 * Received: 0.000000 (double)
FAIL test_readstat (exit status: 1)

Bisecting on an s390x Debian porterbox lead me to commit b8e708d (PR #293). Unfortunately I do not know the file format to debug what is going on.

[1] https://buildd.debian.org/status/fetch.php?pkg=readstat&arch=s390x&ver=1.1.9-1&stamp=1686814711&raw=0
[2] https://buildd.debian.org/status/fetch.php?pkg=readstat&arch=hppa&ver=1.1.9-1&stamp=1686815144&raw=0
[3] https://buildd.debian.org/status/fetch.php?pkg=readstat&arch=powerpc&ver=1.1.9-1&stamp=1686814802&raw=0
[4] https://buildd.debian.org/status/fetch.php?pkg=readstat&arch=ppc64&ver=1.1.9-1&stamp=1686814738&raw=0
[5] https://buildd.debian.org/status/fetch.php?pkg=readstat&arch=sparc64&ver=1.1.9-1&stamp=1686816438&raw=0

[evanmiller]

CC @gorcha

[evanmiller]

I think I see the issue; this only affects the test suite, will merge a fix soon.

[pinotree]

Hm I just tried on a s390x Debian porterbox at commit 0034c8e, and the test still fails in the same way as mentioned in my previous comment.

[evanmiller]

Try 29aac3d

If that doesn't work I'll see if I can emulate big-endian somehow

[pinotree]

Try 29aac3d

PASS: test_readstat

Thanks!