Dirk Eddelbuettel a écrit : > severity 456898 wishlist > tags 456898 + upstream > tags 456898 + wontfix > thanks > > On 19 December 2007 at 11:53, Brian Gough wrote: > | At Tue, 18 Dec 2007 09:23:30 -0600, > | Dirk Eddelbuettel wrote: > | > > /usr/lib/libgsl.so: undefined reference to `cblas_zher2k' > | > > /usr/lib/libgsl.so: undefined reference to `cblas_strsv' > | > > /usr/lib/libgsl.so: undefined reference to `cblas_zdotc_sub' > | > [ many more of these ] > | > > | > Can you see a way forward on this? Is is maybe a question of link order, > ie > | > could -lgslcblas -lgsl be an answer? > | > | The option -Wl,--as-needed needs to be disabled in some way, if it's > | applied globally it's certainly not compatible with libraries that > | depend on other libraries (surely a fairly common occurrence?). The > | link order is not a factor. > | > | Maybe there is some official way to turn off that feature in qmake, or > | to make it behave more intelligently with regard to external libraries > | vs Qt libraries. Otherwise one could put the gsl library at the end > | of the libraries list with -Wl,--no-as-needed: > | > | .... -Wl,--no-as-needed -lgsl -lgslcblas -lm > > As maintainer, I think I concur. GSL made an upstream decision to keep libgsl > and libgslcblas in two libraries --- IIRC this allows to eg substitute > optimised Blas like Atlas or Goto. > > I don't see how qmake can impose an upstream redesign on other libraries, and
qmake doesn't impose anything new. This is already imposed by the Debian Policy for a long time (section 10.2): | Although not enforced by the build tools, shared libraries must be | linked against all libraries that they use symbols from in the same | way that binaries are. This ensures the correct functioning of the | shlibs system and guarantees that all libraries can be safely opened | with dlopen(). Packagers may wish to use the gcc option -Wl,-z,defs | when building a shared library. Since this option enforces symbol | resolution at build time, a missing library reference will be caught | early as a fatal build error. > I do not plan on implmenting one either as I value consistency with upstream > much higher (and would have no time anyway...) As Brian suggests, there also > seems to be a workaround. > > I reset bug levels and tags accordingly. > > Thanks, Dirk > > | > | -- > | Brian Gough > | > -- .''`. Aurelien Jarno | GPG: 1024D/F1BCDB73 : :' : Debian developer | Electrical Engineer `. `' [EMAIL PROTECTED] | [EMAIL PROTECTED] `- people.debian.org/~aurel32 | www.aurel32.net -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
