UMFPACK: avoid warning with SunStudio compiler

[jedbrown]

"../Source/umfpack_qsymbolic.c", line 2693: warning: argument #7 is incompatible with prototype:
        prototype: pointer to function(int, int, int, pointer to int, pointer to int, pointer to int, pointer to void, pointer to double) returning int : "../Source/umfpack_qsymbolic.c", line
 660
        argument : pointer to void

This is a very simple change. Let me know if you need a CLA.

[paulmenzel]

It looks like you have to print a text file, sign it and email it.

[DrTimothyAldenDavis]

I'm constrained by my organizational policy to require a signed contributor agreement before I can merge any PR, even minor ones.

[jedbrown]

An issue with CLAs is that people with blanket authorization to affirm the DCO on open source contributions usually have to route CLAs through institutional legal offices, especially if those CLAs use any non-standard wording. I encourage reading Richard Fontana's article on CLAs. Note that inbound=outbound is explicit in GitHub's TOS.

[DrTimothyAldenDavis]

I understand, but I can't change the policy on my side, and inbound=outbound won't work. Is there a set of canonical CLAs? If mine has non-standard wording, perhaps I could ask to use a standard CLA with the same intent as the one I'm currently constrained to use.

[jedbrown]

You could accept contributions inbound as MIT/Apache-2.0, which is compatible with all your outbound licenses. Yes, in theory, someone could extract specific code lines (that had been merged under these terms) for use in a more permissive distribution, but I think it would be easier for someone with the skills to assemble a "competitor" (filling in the >90% that is not licensed in that way) to write the whole thing from scratch (as Jack Poulson did with Catamari). I understand if you can't/don't want to consider this.