Boost regression notification (2005-10-29 [RC_1_33_0])
Boost regression test failures ------------------------------ Report time: 2005-10-29T15:30:18Z This report lists all regression test failures on release platforms. Detailed report: http://engineering.meta-comm.com/boost-regression/CVS-RC_1_33_0/developer/is... 11 failures in 4 libraries: date_time (1) parameter (1) range (3) serialization (6) |date_time| testtz_database: intel-win32-8_1 |parameter| tutorial: gcc-2.95.3-linux |range| array: vc-6_5 vc-6_5-stlport vc-7_0 |serialization| test_class_info_load_binary_archive: cw-9_5-darwin test_class_info_load_binary_archive_dll: cw-9_5-darwin test_class_info_load_xml_archive: cw-9_5-darwin test_class_info_load_xml_archive_dll: cw-9_5-darwin test_demo_xml_load: intel-win32-8_1 test_shared_ptr_132_text_warchive: intel-win32-8_1
On Sat, 29 Oct 2005 12:00:57 -0500, Douglas Gregor wrote
Looking at the error here, it looks more like an environmental issue than a code issue -- maybe out of space? The code change I put in the other day surely shouldn't be causing an error only here, so I don't think it's that....anyway, here's hoping the next run will clear it up... Jeff
Jeff Garland wrote:
This is once again cause by intel's buggy debug symbol generation in this compiler version. I marked it as expected. (Another solution would be to turn of debug symbols for this test like for some in graph/spirit/...) Stefan
On Sun, 30 Oct 2005 19:20:54 +0100, Stefan Slapeta wrote
I'd be fine with turning off the sysbols although I suppose the risk is users might encounter this compiler bug -- so perhaps it's better to leave the mark up. I presume the best solution to this case is to upgrade to Intel 9.x. Jeff
Jeff Garland wrote:
Yes, and the code to turn off these symbols for intel 8.1 is now already duplicated in about 4 jamfiles :-( Nevertheless, the problems are solved in 9.0. (To be correct: if the symbols grow _very_ large [like e.g. in boost.graph], the problems are the same, but this is because of a restriction in MS obj file format; however, intel 9.0 prints a correct error msg in this case, and it also seems to me that they must have done something with the size of generated debug symbols as those problems occur much less frequently then with 8.1!) Stefan
participants (3)
-
Douglas Gregor -
Jeff Garland -
Stefan Slapeta