Bug 1294201

It is becoming more and more common to encounter very large sparse matrices whose dimensions can exceed the maximum value fitting in a 32-bit integer. For this reason, some projects have switched to an ILP64 BLAS. OpenBLAS in Fedora has shipped it for some time (Bug 1088256), and has just started shipping an alternative ILP64 version with all symbols suffixed with "64_" in order to allow a process to also load libraries built against an LP64 BLAS without conflicts (Bug 1287541).

As I've just switched my Julia package to the ILP64 OpenBLAS, I'd like to be able to use an ARPACK version based on ILP64 OpenBLAS too. Would you be willing to create an alternative version of ARPACK packages for that? Julia uses the following Makefile instructions to use the "64_"-suffixed functions):

ARPACK_OPENBLASFCNS1 := axpy copy gemv geqr2 lacpy lahqr lanhs larnv lartg lascl laset scal trevc trmm trsen gbmv gbtrf gbtrs gttrf gttrs pttrf pttrs
ARPACK_OPENBLASFCNS2 := dot ger labad laev2 lamch lanst lanv2 lapy2 larf larfg lasr nrm2 orm2r rot steqr swap
ARPACK_OPENBLASFCNS3 := dotc geru unm2r
ARPACK_OPENBLASFCNS4 := COPY LABAD LAMCH LANHS LANV2 LARFG ROT GEMV
ARPACK_FFLAGS += $(foreach fcn, $(ARPACK_OPENBLASFCNS1) $(ARPACK_OPENBLASFCNS2), -Ds$(fcn)=s$(fcn)_64 -Dd$(fcn)=d$(fcn)_64)
ARPACK_FFLAGS += $(foreach fcn, $(ARPACK_OPENBLASFCNS1) $(ARPACK_OPENBLASFCNS3), -Dc$(fcn)=c$(fcn)_64 -Dz$(fcn)=z$(fcn)_64)
ARPACK_FFLAGS += $(foreach fcn, $(ARPACK_OPENBLASFCNS4), -DS$(fcn)=S$(fcn)_64 -DD$(fcn)=D$(fcn)_64)
ARPACK_FFLAGS += -Dscnrm2=scnrm2_64 -Ddznrm2=dznrm2_64 -Dcsscal=csscal_64 -Dzdscal=zdscal_64
ARPACK_CFLAGS += -Dsgemm_=sgemm_64_ -Dcheev_=cheev_64_