Boost logo

Boost Testing :

From: Martin Wille (mw8329_at_[hidden])
Date: 2005-05-09 01:35:27

Hartmut Kaiser wrote:

> Could you please clean out the current state and force a full CVS update?

The code always gets updated completely from the CVS.

Since this is failing on so many platforms, this might be a good chance
to find out what is wrong with the build system or the test system here.

The situation here is:
The testwave binary created by the most recent run shows a creation date
of yesterday and the test_log.xml output is newer than the binary.
However, the testwave.o file used for creating the binary is older.
The directory listing for gcc 3.4.3 is (times are shown in UTC):

-rw-r--r-- 1 m m 807 May 8 09:52 test_log.xml
-rwx--x--x 1 m m 64626929 May 8 09:21 testwave
-rw-r--r-- 1 m m 890820 May 7 16:57 testwave.o
-rw-r--r-- 1 m m 287 May 8 09:51 testwave.output
-rw-r--r-- 1 m m 9112748 May 8 09:20 testwave_app.o

The listing for the wave library directory is:

-rw-r--r-- 1 m m 61751022 Apr 22 23:51 libboost_wave-gcc-sd-1_32.a
-rw-r--r-- 1 m m 64106376 May 8 07:59 libboost_wave-gcc-sd-1_33.a
-rw-r--r-- 1 m m 444 May 8 09:52 test_log.xml

AHA! Does Wave still pick up the 1.32 version of the library?

test_log.xml for gcc.3.4.3 from yesterday reads:

<test-log library="wave" test-name="testwave" test-type="run"
toolset="gcc-3.4.3-linux" show-run-output="false">
<link result="succeed" timestamp="2005-05-08 09:52:16 UTC">
<run result="fail" timestamp="2005-05-08 09:52:24 UTC">
#line 24 "t_6_059.cpp"

expected error:
t_6_059.cpp(21): warning: character literal out of range: L'\xabcdef012'

testwave: 168 of 169 test(s) succeeded (1 test(s) failed).


This looks exactly like what gets displayed on the metacomm site.

However, when I run the test manually (using yesterday's binary) then I
get this output:

./testwave --config-file
exception caught: unknown option long_long
testwave: 166 of 169 test(s) succeeded (3 test(s) failed).

The exception gets thrown when testfiles/t_9_014.cpp is being processed.
The status code returned is 3.

testwave is statically linked, so this can't be an issue with a wrong
shared library getting used.

I have no idea what is happening here. Even if the 1.32 version of
libboost_wave would get used, the results of the automated and the
manual runs should still be consistent.


Boost-testing list run by mbergal at