Move deprecation

[colinleach]

I'm not sure whether we should merge this, and certainly not today. I experimented with deprecating old-annalyns-infiltration and its UUID at the top level. It didn't go entirely smoothly: the linter insists on the full exercise files being in exercises/concept even for deprecated exercises.

Maybe worth a gamble to untangle things, if we have no better ideas?

Along the way, I found out why Lasagna works: both old and new versions have the same UUID. I think that means we could delete the one in concept.wip/????

[depial]

I'm not sure whether we should merge this, and certainly not today.

I'm okay on waiting on this for now.

Along the way, I found out why Lasagna works: both old and new versions have the same UUID.

This is interesting. I can't see why deleting how deleting one would break things, but I'm not really in a good position to say ;) From now on, should we try just duplicating the UUIDs when we use a preexisting concept instead of creating new ones? Maybe this is the expected procedure?

In other news, the website doesn't seem to have changed overnight...

[colinleach]

Landing page of the website just now:

It looks like swapping UUIDs for the conditionals concept had regrettable side effects: the website apparently never forgets a UUID once it's seen it. Hence 2 concept boxes on the tree. I just checked the Exercism documentation, and it doesn't look like a concept entry can have a "status" field we can use to deprecate it - that's just for exercises.

My guess is that we can (probably) fix this by recycling the unwanted UUID in the next new concept (e.g. pairs-and-dicts).

I'll drop this PR to draft, for now. We can decide later whether to merge it or just close it.

Does that make sense?

[colinleach]

On a related subject, these are the UUIDs associated with elyses-enchantments:

"185683d8-96f7-4d51-8726-56e18b6ef0c7" (top level)
"234466a0-08ce-45a1-aebc-5ec0b0cde24f" (in concept.wip)

If getting rid of a conditionals duplicate works as we hope, I'd like to do the same with the "234466a0-08ce-45a1-aebc-5ec0b0cde24f" UUID.

From now on, should we try just duplicating the UUIDs when we use a preexisting concept instead of creating new ones?

Mostly, this is what I've been doing (plus removing the original exercise files). A couple of the early ones got away from me, when I started working on the legacy stuff last September. We're still cleaning up the mess that caused.

[depial]

Landing page of the website just now

At least it doesn't show up in the syllabus, which is more than I can say for the two Booleans... :D

Does that make sense?

Yep! (at least I hope so)

[depial]

If we have to move it to deprecate it, I suppose we should try.

I'm only worried that we'll somehow end up with another duplicate somehow...

Another possible action: Could we try deprecating an existing exercise in exercises\concept, then return it to wip? Or is deprecation a point of no return? If it's a switch we can throw, we could at least test the effects beforehand, but, at the same time, I hesitate to mess with something that's not broken :D

[colinleach]

is deprecation a point of no return?

I don't know, but I think it might be.

I hesitate to mess with something that's not broken

Agreed! This would make me very nervous.

we'll somehow end up with another duplicate

Perfectly understandable paranoia (given recent experiences), but I don't think this is very likely. Either way, we end up having to ask somebody more senior (probably Jeremy) to sort it out for us.

[colinleach]

Success!

[depial]

Success!

I think it just reset the exercise, so we'll have to solve it again, but it looks good so far!

[colinleach]

Yes, I just solved it again and it's looking good.

That just leaves Vectors. I don't have any good suggestions for that. Is it time for another forum post, to get the other maintainers involved?

[depial]

Is it time for another forum post, to get the other maintainers involved?

Possibly :)

I know there is a preexisting Conditionals concept, but was there something else strange with Elyse's Enchantments? I feel like the former is what is keeping the concept from linking, while the latter could be what is resulting 0 exercises appearing. Of course, it could all be down to the Conditionals confusion.

Do we have any methods at our disposal?

[depial]

I'm fairly convince it's got to be the Conditionals duplicate. Could we reuse the UUID of the old concept with a new concept, and delete the old conditionals entry in the concepts.wip\config.json?

I know there is no documentation about deprecating concepts, but that's also one thing we could ask about. Maybe it's possible?

[colinleach]

I've no easy answers and need time (and energy) to think about this. Right now, I'm confused about which UUIDs are already recycled, which are duplicated (e,g, Lasagna). I need to sort this out in my head.

Could we reuse the UUID of the old concept with a new concept, and delete the old conditionals entry in the concepts.wip\config.json?

We'll probably end up doing this.

[depial]

We'll probably end up doing this.

I figure this is our best shot at fixing things. Maybe we could try renaming the dupe? I figure if we try to stick with what just worked for Annalyn's Infiltration, we might have a shot.

Of course, it wouldn't hurt to ask other maintainers :)

[colinleach]

After a bit more thought (probably not enough):

• I'm in favor of reusing the UUID of the conditionals dupe. I don't think renaming the slug will be enough - we hit success with annelyns-infiltration after moving to exercises/config and marking deprecated, which (I think) only works for exercises.
• We could repeat that deprecation trick with the elyses-enchantments dupe. That still has its own UUID.
• In concept.wip, both lasagna and old-annalyns-infiltration are now redundant. We already reused those UUIDs at the top level, so the dupe should be deleted. Should I PR that?
• I'll check the unmerged PRs to see if there are any we can use to recycle UUIDs. My first thought was functions, which is the next one on my list, but that's already committed to a functions dupe.
• I'll do a more systematic search for what UUIDs are used where. Old habits die hard, and I still like having these sort of lists in PostgreSQL for searching (overkill, obviously, but it's harmless and makes an old man happy).

[colinleach]

Also, I'll post an update on the forum. Some people might be interested, and (other than Jeremy and his international peregrinations) it may be easier to get attention before Bootcamp pert 2 starts April 26.

[depial]

We could repeat that deprecation trick with the elyses-enchantments dupe.

That seems like it could be beneficial (and I assume neutral a worst).

Should I PR that?

At present, I don't see a problem with deleting these if they are using the same UUIDs as the top-level exercises.

Also, I'll post an update on the forum.

Sounds good. Try to see if it's possible to deprecate a concept. If not, I'm starting to think we'll need help sorting out the Vectors / Conditionals connection issue.