Preliminary implementation of default values

[nshaffer]

In support of #62

• New module "default_m" which exports the generic function name "default"
• New entries in src/CMakeLists.txt and src/Makefile.manual

I've implemented a generic interface for real(sp), real(dp), real(qp), int16, int32, int64, logical, and character. I figure this is enough for a first pass. Complex types and discriminating between ASCII and UCS characters can come later. Probably no need to do specific logical kinds.

What's missing right now is tests. I don't know how to use CTest, but I would like to learn if anyone can link me a decent tutorial.

[jvdp1]

Nice addition. Thanks. My comments are mainly associated with other PR or unresolved issues.

[certik]

Thanks @nshaffer. I resolved the conflict with master and committed to your branch. Make sure you pull it before adding more commits.

Also, I recommend people to submit PRs from a different branch than their master, so that they can use their master to track the upstream master (to rebase, merge, etc.).

[certik]

Regarding the test, try to copy let's say the test_ascii test and its CMake setup. It should be straightforward. If you have any questions, let us know.

[nshaffer]

Thanks all for the comments. Seems like folks in #62 prefer the name "optval" so I converted to that everywhere and renamed the fallback argument "default", so that you can write optval(x, default=1.0) if you want to be explicit. Also wrote tests for all currently implemented types.

[certik]

Thank you! This is great. I'll review later tonight or tomorrow.

…

On Fri, Jan 3, 2020, at 5:59 PM, nshaffer wrote: Thanks all for the comments. Seems like folks in #62 <#62> prefer the name "optval" so I converted to that everywhere and renamed the fallback argument "default", so that you can write either `optval(x, default=1.0)` if you want to be explicit. Also wrote tests for all currently implemented types. — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#73?email_source=notifications&email_token=AAAFAWAYRYK3SLJO6IHLFUDQ37NQPA5CNFSM4KCPS462YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGOEICNLBI#issuecomment-570742149>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAAFAWDH262SE7MZ3L5F6OLQ37NQPANCNFSM4KCPS46Q>.

[certik]

This looks great. The API has the right order of arguments that everybody agrees, and optval seems to be the name that most also agree on. There are tests, implementation is clean. So let's merge it and start using it. For example in #71.

[certik]

Thanks @nshaffer for the work!

[certik]

Here is the first usage of optval: a8416a1

[nshaffer]

Cheers!

It occurs to me that optval should be pure elemental for the intrinsic types. Shall I tack that on here or open a new PR? (I've added a dev branch to my fork now, if that makes any difference.)

[certik]

Yes, send another PR please against master, from a new branch.