tst_h5_endians fails on s390x

[opoplawski]

When building the Fedora netcdf 4.9.3-rc2 package on rawhide I get the following on s390x:

===============================================
   netCDF 4.9.3-rc2: nc_test4/test-suite.log
===============================================
# TOTAL: 85
# PASS:  84
# SKIP:  0
# XFAIL: 0
# FAIL:  1
# XPASS: 0
# ERROR: 0
System information (uname -a): Linux 6.11.8-300.fc41.s390x #1 SMP Thu Nov 14 20:10:32 UTC 2024 s390x
Distribution information (/etc/os-release):
NAME="Fedora Linux"
VERSION="42 (Rawhide Prerelease)"
RELEASE_TYPE=development
ID=fedora
VERSION_ID=42
VERSION_CODENAME=""
PLATFORM_ID="platform:f42"
PRETTY_NAME="Fedora Linux 42 (Rawhide Prerelease)"
.. contents:: :depth: 2
FAIL: tst_h5_endians
====================
* Checking that endianness is properly read from file.
** Generating test files.
*** Creating a file via netcdf API: tst_h5_endians.nc.
** Checking test files.
*** tst_h5_endians.nc
	Little-Endian Float...	passed
	Big-Endian Float...	passed
	Little-Endian Int...	passed
	Big-Endian Int...	passed
	Little-Endian Double...	passed
	Big-Endian Double...	passed
** Read/Write test files.
*** tst_h5_endians.nc
	Little-Endian Float...	failed
	Little-Endian Int...	failed
	Big-Endian Int...		Little-Endian Double...	failed
** Failures Returned: [3]
FAIL tst_h5_endians (exit status: 3)
```

[opoplawski]

Same as #3284 ?

[edhartnett]

Yes, there's a fix. It's not clear when or if it will be merged.

…

On Wed, Mar 4, 2026 at 8:12 PM Orion Poplawski ***@***.***> wrote: *opoplawski* left a comment (Unidata/netcdf-c#3062) <#3062 (comment)> Same as #3284 <#3284> ? — Reply to this email directly, view it on GitHub <#3062 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ABCSXXAHWV6GJJSJ6EKORBD4PDWDXAVCNFSM6AAAAACWHNG222VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHM2DAMBRHAZDMMRUGU> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>

[WardF]

I will take a look at getting the fix for this test in, while also wiring in the big endian tests via emulation, but I don't want the remaining big endian failures to get lost in the mix; I'll need to get those addressed as well. But I'm working through the backlog now. Thanks!