Fall Challenge 2020 - Bugs & Questions

Solved by doing a warmup calling a lot of streams : streams are the problem.

Now i will migrate to c++ ^^

1 Like

Use simulate() here: https://github.com/CodinGame/FallChallenge2020/blob/main/src/test/java/Fall2020Main.java#L20

You may want to reset the ID counter for spells when you run the simulate in a loop. And you have to create a new GameManager in each iteration.

1 Like

With the updates made recently, timing can still be impacted, but it is manageable. With reduced batch size to account for inconsistencies, my Rust code no longer gets timeouts.

There are bugs always / all times when i launch code it keep saying ā€œyou wrote before reading entries, it may make weird behaviorā€ but itā€™s clearly not and even when i remove the ECHO/PRINT commands it keep saying that :confused:

Are there any memory restrictions for submitted code?

Do you mean usable memory when running? => https://www.codingame.com/faq
768MB of RAMā€¦

For the size of your code (ā€œNumber of linesā€ / ā€œnumber of charsā€): Yes, its 100k bytes.

ā€œnumber of charsā€ != ā€œnumber of bytesā€, so 100k chars code limit could be a bit higher when counted in bytes (around 200 kB).

Can CG do something about Rust (and maybe some other languages) ?

Rust is in production mode in the arena, but in debug mode in IDE. You canā€™t test against Rust opponents in the IDE because their code crash or is at least very bad/slow (intentionally or not).

Nobody can test his bot against emilā€™s bot. Itā€™s absurd ā€¦

Our opponent bot should be in production/release mode in our IDE.

12 Likes

Please read the instructions in the ā€œDebugging tipsā€ :slight_smile:

" * Append text after any command and that text will appear next to your witch"

Cheers

Hi all, I observed a curious thing on my test :
When I learn a spell with action id 37, it becomes the action id 87 in my known spells.
Is it a bug ?
Thanks for your help. I have pictures of my action but I canā€™t share it here because Iā€™m a ā€œnew memberā€ =)

Not a bug or question just a small remark:
I checked league sizes for past 3 contests and projected the average to Fall Challenge current playersize:
image

Based on this, if your personal goal is to finishā€¦

  • in Legend: then you shall aim to be in the top 3% (or top ~130)
  • in Gold: aim to be in top 22% (or top ~750)
  • in Silver: ā€¦ nah, aim a bit higher, you can do it

Data variation is high: Spring challengeā€™s Silver Boss was tougher than usual. Wood 1 Boss of Ocean of Code also.

4 Likes

From the puzzle technical details : After learning a spell, the learnt version of the spell has a different id from the one previously within the tome.

You need to include the current contest in the prediction as silver surpased 1200

Put a guesstimate of 6.3k participants

@Magus, my Rust code runs now in IDE and arena (from the matches Iā€™ve viewed in results). It isnā€™t as good as Iā€™d like, but should work.

Meh ā€¦ as soon as I say this, I get a timeout in IDE. So, make it 99% running fine?

In my table only final silver is set at 1200.
Unrelated to current silver count, as many will go up to gold/legend. And of course many will come up from bronze. 1200 guess is still feasible.

Same here. Really annoying.
Here is an exemple of simple code running same simple actions on every turn in about 3-4msā€¦ and we can see spikes of ~20ms on turn 1-2 and 30ms on turn 18 : https://www.codingame.com/replay/504584619

Thanks a lot ! :wink:
I should have been more attentive !
Good game !

That message is often shown when the real cause is a timeout. Check the last lines of output to be sure.
image

At the end of the game the number of ingredients in your inventory that are of tiers 1-3 is added to your score.

[BUG][DART]
I also had some issue concerning the following subject :

  • [Investigating] Desynchronisation warning after a timeout*

I think I resolved it by limiting the execution time of some part of my code, but I donā€™t really understand what is the maximum execution time per turn because my code spends generally more than 50ms.

In the notice, it is written :
Response time per revolution ā‰¤ 50ms
First turn response time ā‰¤ 1000ms

@TwoSteps Could you give more information on the execution time of our code ?