Bug#957828: sparskit: ftbfs with GCC-10

Étienne Mollier etienne.mollier at mailoo.org
Fri Jan 15 12:16:30 GMT 2021


Control: tag -1 patch

Hi,

I prepared a targeted patch to fix Gcc-10 issues with sparskit,
and ran the embedded ./dotests to make sure it did not introduce
any errors.  This fixes the argument type mismatch in the csrcsc
call by introducing the one dimension vector iziama of type
real(8), instead of an integer(4) scalar, while trying to
maintain compatibility with Fortran 77.

Kind Regards,
-- 
Étienne Mollier <etienne.mollier at mailoo.org>
Fingerprint:  8f91 b227 c7d6 f2b1 948c  8236 793c f67e 8f0d 11da
Sent from /dev/pts/5, please excuse my verbosity.

--- sparskit-2.0.0.orig/ORDERINGS/ccn.f
+++ sparskit-2.0.0/ORDERINGS/ccn.f
@@ -90,7 +90,7 @@
 c                      July 1992 - Update: March 1994
 C-----------------------------------------------------------------------
       integer izs(nw), lpw(n), nsbloc(0:nblcmx), ia(n+1), ja(*)
-      real*8 amat(*)
+      real*8 amat(*), iziama(1)
       logical impr
       character*6 chsubr
 C-----------------------------------------------------------------------
@@ -147,10 +147,12 @@
      *                ja, ia, izs(ilpw), izs(ilpw), job)
            ipos = 1
 c..........We sort columns inside JA.
-           call csrcsc(n, job, ipos, amat, ja, ia, izs(iamat),
+           iziama(1) = izs(iamat)
+           call csrcsc(n, job, ipos, amat, ja, ia, iziama,
      *                 izs(ijat), izs(iiat))
-           call csrcsc(n, job, ipos, izs(iamat), izs(ijat), izs(iiat),
+           call csrcsc(n, job, ipos, iziama, izs(ijat), izs(iiat),
      *                 amat, ja, ia)
+           izs(iamat) = iziama(1)
       endif
 c.....We modify the ordering of unknowns in LPW
       call compos(n, lpw, izs(ilpw))
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 833 bytes
Desc: not available
URL: <http://alioth-lists.debian.net/pipermail/debian-science-maintainers/attachments/20210115/38b33905/attachment.sig>


More information about the debian-science-maintainers mailing list