Update docstring of simDeath() in core.

[pkofod]

Docstring doesn't match behavior.

[mnwhite]

Good catch! But my typo is in the opposite direction: the docstring says what it's supposed to do, but the np.ones is supposed to be an np.zeros. We can also consider removing that print statement and letting "no death and replacement" be the default.

[pkofod]

But my typo is in the opposite direction: the docstring says what it's supposed to do, but the np.ones is supposed to be an np.zeros.

I thought that might be the case, but I couldn't determine if you intended for everyone to die by default and thereby force people to respect the printed message, or if it was a typo the other way. In that case, I will change the behavior! (and add a test), and remove the printed message. Then noone dies, unless the users specifies it.

[llorracc]

Sounds like "what to do" is now agreed on this? Or is more discussion needed?

[mnwhite]

I think Patrick already did it. Let me take a look tonight and see if it's ready to merge (probably is).

…

On Sat, Jun 15, 2019 at 4:38 PM Christopher Llorracc Carroll < ***@***.***> wrote: Sounds like "what to do" is now agreed on this? Or is more discussion needed? — You are receiving this because you commented. Reply to this email directly, view it on GitHub <#302?email_source=notifications&email_token=ADKRAFNEZ27AVP7HG73TXKLP2VHNZA5CNFSM4HXZKEY2YY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODXY7W4Q#issuecomment-502397810>, or mute the thread <https://github.com/notifications/unsubscribe-auth/ADKRAFMX7ND4UNWGWCRNXSDP2VHNZANCNFSM4HXZKEYQ> .

[pkofod]

I removed the warning, and changed the default.

[shaunagm]

@pkofod, can you write a release note for this?

[pkofod]

"Fix unintended behavior in default simDeath(). Previously, all agents would die off in the first period, but they were meant to always survive."

This change is technically breaking. Before, there would be a warning that the users should define their own method and all agents would die. Now we simply assume that agents live forever, and people have to define a method to avoid this. That said, bug fixes (this had unintended behavior) are not always interpreted as requiring a major version bump under semver 2.0, especially so if no users are expected to rely on the old (wrong) behavior. Since the old method was accompanied by a warning to define a new method, I think we're in the clear.