Fix Dgels to check the correct dimension

[btracey]

No description provided.

[kortschak]

Are you sure? Netlib defines MN = MIN( M, N ).

[btracey]

It does? From Netlib Dgels: "The leading dimension of the array B. LDB >= MAX(1,M,N)."

B is an M x NRHS matrix. X is an N x NRHS matrix. Both of these matrices have to fit into B, so the number of rows has to be at least max(m,n)

[kortschak]

The issue is that NETLIB makes the check of ldb against max(m, n), but uses mn = min(m, n) for checking against lwork.

      INFO = 0
      MN = MIN( M, N )
      LQUERY = ( LWORK.EQ.-1 )
      IF( .NOT.( LSAME( TRANS, 'N' ) .OR. LSAME( TRANS, 'T' ) ) ) THEN
         INFO = -1
      ELSE IF( M.LT.0 ) THEN
         INFO = -2
      ELSE IF( N.LT.0 ) THEN
         INFO = -3
      ELSE IF( NRHS.LT.0 ) THEN
         INFO = -4
      ELSE IF( LDA.LT.MAX( 1, M ) ) THEN
         INFO = -6
      ELSE IF( LDB.LT.MAX( 1, M, N ) ) THEN
         INFO = -8
      ELSE IF( LWORK.LT.MAX( 1, MN+MAX( MN, NRHS ) ) .AND. .NOT.LQUERY ) THEN
         INFO = -10
      END IF

[kortschak]

It looks to me like the real fix is to change the line below (no comment available there) to checkMatrix(max(m, n), nrhs, b, ldb), and revert this.

[btracey]

Yep. You're totally right. Sorry. Fixed.

[btracey]

Again, I don't understand, one fails on 1.4, the other on 1.5rc1.

[btracey]

Ah, I see what you mean by "real failure". It's a time out test failure, and doesn't seem to be consistent as to which test it fails on. Maybe we should increase the test timeout to 20minutes?

[kortschak]

What concerns me is the fact that there is such a discrepancy in the amount of time it takes to complete cgo tests on travis compared to a local machine - it's seconds on my laptop and ~ten minutes on travis. This really indicates to me something is not right.

[kortschak]

LGTM

[btracey]

I agree that it's weird. It seems like it has to be a cgo problem or a travis problem. If it's a travis problem there's nothing we can do about it. I don't know how to test if it's a cgo problem. We're running on different operating systems and it's fine 100% of the time (so far) on my machine.

[kortschak]

I think Seb's answer is probably the reality.

[btracey]

Seb's answer?

[kortschak]

https://groups.google.com/d/topic/golang-nuts/MYBwEq2Sg_c/discussion