Re: [boost] Boost regression notification (2006-01-06 [HEAD])

From: Daryle Walker (darylew_at_[hidden])
Date: 2006-01-06 19:17:32

On 1/6/06 12:00 PM, "Douglas Gregor" <dgregor_at_[hidden]> wrote:

> Boost Regression test failures
> Report time: 2006-01-06T14:16:52Z
> 
> This report lists all regression test failures on release platforms.
> 
> Detailed report:
> http://engineering.meta-comm.com/boost-regression/CVS-HEAD/developer/issues.ht
> ml
[SNIP]

The version of this page I'm looking at is marked: "Report Time: Fri, 6 Jan
2006 21:16:27 +0000".

> |rational|
> rational_test:  gcc-2.95.3-stlport-4.6.2-linux  intel-9.0-linux
> intel-win32-9_0  vc-7_1  vc-7_1
[TRUNCATE]

gcc-2.95.3-stlport-4.6.2-linux:
    Does GCC 2.95.3 have ADL issues?  For the GCD and LCM tests, "using
boost::gcd" statements are made and the simple name is used for the actual
calls.  Those functions can't be found, but _only_ for the custom type,
which is in an anonymous namespace.  No errors are given for the versions of
the test template that use built-in types.

intel-9.0-linux:
    Yesterday's regression report had this entry on its web page point to an
invalid page notice.  Today's report doesn't even mention this compiler at
all!  Whatever sync/upload problems you have need to be fixed.

intel-win32-9_0:
    I'm seeing the same problems as yesterday, including the printout of the
problem code.  Unfortunately, I've radically changed that code, and I see
the changed code in the web-CVS.  This test system needs to do a CVS update
_before_ the next test cycle (or clear its caches).

vc-7_1:
    This system is listed twice here, but four times on the web page.  The
web page has two different VC 7.1 test systems, with two entries each.  The
"Bronek_office" system can't link to the Boost.Unit-Test library.  The
"RudbekAssociates" system complains that "operator new" is defined too many
times.  This latter problem was brought up in a different thread on this
list, where it was instigated by the Boost.Test library.  I think the author
said in that thread that it shouldn't be a problem.  (Maybe the first system
has the same problem, but it crapped out before telling us.)  In either
case, it seems that VC 7.1 likes the actual rational test code; it just
chokes at the end due to linking problems.

-- 
Daryle Walker
Mac, Internet, and Video Game Junkie
darylew AT hotmail DOT com