В Втр, 11/03/2008 в 10:15 +0100, Frank Strauß пишет:
Hi Peter,
Am 03.03.2008 um 06:23 schrieb Peter Volkov:
[snip] comparing `smidump -f smiv2 SNMPv2-MIB' output with dumps/orig-smiv2/ *. comparing `smidump -f smiv2 IF-MIB' output with dumps/orig-smiv2/*. comparing `smidump -f smiv2 MAU-MIB' output with dumps/orig-smiv2/*. comparing `smidump -f smiv2 RMON2-MIB' output with dumps/orig-smiv2/*. *** smidump output differs, see smidump-orig-smiv2.out/*.diff FAIL: smidump-orig-smiv2.test
Since I could not reproduce this test case failure either, I looked at your supplied test output files from the "orig-smiv2" test case. There was only a difference in the order of imported items. This does NOT affect the correctness of the emited files.
However, it would be interesting to know how this happened. I can just guess that it's something like a differing order of results from readdir(3) calls when MIBs are looked up, or something similar. I admit that my enthusiasm is limited to get this to a more deterministic fashion.
Oh, seems that this due to different locale settings. I've tried with LC_COLLATE=C and then only sming failed. Now I forced LC_ALL=C in ebuild, but may be it's better to force that in test suite. If there will be any update from hppa team I'll mail you. Thank you!