[$40] Update PATCH timeline endpoint to update dates of milestones

[vikasrohit]

When user is updating the start date of a timeline using the PATCH /v4/timeline/123 endpoint, it should also update dates(startDate and endDate) of all milestones for that timeline. It should start the first milestone (identified via order=1) with timelines start date and calculate end date of that milestone using duration and update the subsequent milestones in the same way.

fyi @coderReview

[architectt1]

@coderReview @vikasrohit Are milestones of a given timeline supposed to be one after the other? for example:

• Milestone 1: day 1-3.
• Milestone 2: day 4-7.
• Milestone 3: day 8-10.

And given this example, would change the startDate of the timeline to day 5 make the above:

• Milestone 1: day 5-7.
• Milestone 2: day 8-11.
• Milestone 3: day 12-14.

?

(I know these are actually dates, but I'm putting an easy-to-follow example with numbers).

[architectt1]

Answering to myself, I believe this is correct, as there's some code doing exactly that when creating milestones from ProductMilestoneTemplate records.

[vikasrohit]

Correct. Our backend should always ensure that milestones dates are in sync as per their order.

[architectt1]

Got it. One thing I'm going to rely on is in that the data is correct in the database. For example there might be milestones that step into each other. I.e.

• Milestone 1: day 5-7.
• Milestone 2: day 6-8.
• Milestone 3: day 7-10.

For all cases I'm just going to add the timeline difference to the start date, and then set the end date to be milestone start date + milestone duration. I believe this is a fair approach to use.

[architectt1]

Hmm actually because of a test I saw, even if a milestone has a startDate and a duration, it can have an end date not set. Apparently this should stay as that, so in those scenarios I'll just update the startDate.

[vikasrohit]

That case should not happen ever. We should assume that database state is correct i.e. milestones are strictly after one another. And always follow the simple rule of setting the dates strictly after one another.
To avoid inconsistent dates in database, I think we need to make sure that PATCH /v4/timelines/123/milestones/1234 does not allow setting it incorrect. May be we can handle that in #127

[architectt1]

Do you mean the dates stepping into each other, or the null endDate, or both? the endDate column is marked as nullable in the database. Maybe I should always set the endDate, even when null (= startDate + duration)?

[vikasrohit]

Right now I can not collect any use case of having null endDate, Yes please, set endDate even if it is null.

[architectt1]

#139