Joaquín Mª López Muñoz writes:
Aleksey Gurtovoy ha escrito:
Joaquín Mª López Muñoz writes:
The attached fix makes all tests compile for msvc (not all run, though) and hopefully also for the other two toolsets (can't try myself.)
Joaquín,
Could you please verify that you patch was correctly applied? It looks like things got better with MSVC in terms of compilation, but now "list_inserter" and "list_of" tests fail at run time -- http://tinyurl.com/6g499 ("multi_index_container" doesn't fail, but gets terminated by regression tools because apparently it takes more than 5 minutes to run).
I think the patch is correctly applied. When I submitted it, I reported that the tests list_inserter, list_of and multi_index_container would still crash at run-time,
Oh, I missed this somehow.
which is what's happening (more or less):
* test_list_inserter: behaving as predicted * list_of: behaving as predicted (in fact, this test does not depend on the patch at all, list_inserter.hpp is not included) * multi_index_container: here we have differing results; in my local environment, there's no hang up, but a sure crash.
OK. Thorsten, could you tell us what's up with the crashes?
Now I wonder about the results of test array for msvc-stlport and vc7. This test does not depend on the patch I proposed, so its current failing on these platforms must be due to something else. Does anybody remember whether they used to work in past days?
They were failing before, see http://thread.gmane.org/gmane.comp.lib.boost.devel/110081. -- Aleksey Gurtovoy MetaCommunications Engineering