Summary: | sci-mathematics/flint USE=ntl - src_test(): x86_64-pc-linux-gnu-g++: error: build/interfaces/NTL-interface.o: No such file or directory | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | François Bissey <frp.bissey> |
Component: | Current packages | Assignee: | Patrick Lauer <patrick> |
Status: | RESOLVED FIXED | ||
Severity: | normal | CC: | frp.bissey, hasufell |
Priority: | Normal | Keywords: | PATCH |
Version: | unspecified | ||
Hardware: | All | ||
OS: | Linux | ||
See Also: | https://github.com/wbhart/flint2/issues/72 | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- |
Description
François Bissey
2014-07-02 01:50:28 UTC
(In reply to Francois Bissey from comment #0) > I notice however that Julian missed a problem we had fixed in the overlay: Has it been reported upstream? + 02 Jul 2014; Julian Ospald <hasufell@gentoo.org> flint-2.4.4.ebuild, + +files/flint-2.4.4-test.patch: + fix NTL test build wrt #516028 No. The story on why it hasn't is that you can't reproduce it on a vanilla build or one made inside sage. That puzzled me for some time but I think I finally worked out what is happening (after your prompting about upstream really). By default as delivered by upstream flint build both static and shared libraries and objects. So in a vanilla version of flint you have both .lo and .o objects around and this is not an issue. In Gentoo by default we turn off the static part of the build, exit the .o file. Upstream may be willing to switch the ntl test to .lo, an educated guess right now without looking at the code is that the rest of the tests are using .lo - that or the ntl test is different from the rest. |