Languages Update Issues


#1

Hello,

we're currently updating the languages (and in particular C++). This seems to cause a certain number of timeouts.
We'll investigate it as soon as possible and keep you updated. I'll share the complete list of updated versions tomorrow once it's completed.


C# is broken today
#2

C++ seems to be now in C++14.
Some of my multiplayers puzzles codes won't compile anymore :smiley: Well, no big changes but it's weird.

I guess codingame server compile the code every time for every games in the arena ? We could win a lot of times on submits without this compilation phase :stuck_out_tongue: (compiling a C++ code can take a few seconds. Same in Java. Few seconds multiply by many games it's a lot).

I need to check all multiplayers puzzles for compilation error.

But the weird of all is Smash the Code. My code compile but i have random timeout against random players. And most of the time it's the opponent who got a timeout ... :confused:


Timeouts
#3

Hello @Magus,

Regarding timeouts, do they occur during a play or when browsing last battles after a submit? If a last battle could you send the replay link?

Thanks.


#4

It was during a play in the IDE. I was testing who got timeouts or not. I didn't submit anything at Smash the code yesterday.


#5

I could reproduce this issue for some play battles that would timeout on the first frame. When I can reproduce, I can reproduce always by freezing the seed.

Maybe the startup of plays in C++ (that go through gdb) is slightly slower with the new version of g++/gdb. If that's the case I suspect gdb (which, as a reminder is not used when submitting in the arena). Another explanation would be that players like FredericBautista are aware that there code may fail during plays but don't fail when in the arena due to the removal of gdb.

Anyway if you have issues with plays not on the first frame, please let me know and give me the seed and exact players.


#6

I had the same this morning, several timeouts in STC IDE but not as recurring in the arena though...


#7

It was always at the first frame.
I'll try again to give you players and seeds.

For example i tested against Yakalir. We don't timeout that in the arena. But if i play against him in the IDE, one of us will always timeout at the first frame. Always. No matter the seend.

But if i play against reCurse, none of us timeout.


#8

I try to complete 2 community puzzles in C# today and see a problem.
"PLAY ALL TESTCASES" command gives me 100% success.
Clicking "SUBMIT", I see 0% success: all testcases give an errors.
I sure my solutions are valid.

And, I did one observation: there is the next "Console Output" when I make an syntax error in the code:

Console output
Errors
'/tmp/deps' in MONO_PATH doesn't exist or has wrong permissions.
/tmp/Answer.cs(21,8): error CS1525: Unexpected symbol `.'
Compilation failed: 1 error(s), 0 warnings

Take attention to the first line of the output with '/tmp/deps'...
Please, locate the problem and repair it.


#9

@yizraor I just moved your topic here since it's a direct consequence of the update. It seems to happen for all languages linked to .net

We are currently working on fixing this. Not sure we'll be able to fix it before the weekend though :frowning:


#10

The issue which caused 0% validation with C#, F# or VB.NET should be solved.


#12

After the update from python 2.7.13 to 2.7.14 the tensorflow puzzle is broken:

ImportError: No module named tensorflow.python.platform
at input_data.py. in <module> on line 21
at Answer.py. in <module> on line 6

I have a solution, that was working before the update, but now gets the above error message.


#13

Ok, I guess we need to reinstall the tensorflow module. Will be done on Monday. Thanks for spotting the problem.


#14

Hi,
I've got some issues with C#.
Since 3pm on "La Bataille" game, tests fail with error "'/tmp/deps' in MONO_PATH doesn't exist or has wrong permissions."
I think it's the same issue as yizraor.

Thanks for the help.


#15

hey @Sethangel79, if I remember well what @_CG_XorMode told me, this console issue does not impact anything. The main issue that @yizraor had was the validation issue while tests passed in IDE. We'll fix this issue anyway but I think your tests fail for an another reason.

EDIT: the lines below are regular compilation issues :wink:


#16

Hi and thank you for your answer.
I will think about twice the next time before bother you.
I apologize and i'm going to check my code :blush:


#17

no worries :wink:


#18

The MONO_PATH error message is fixed and no longer appears on C# compilation failure.


#19

For example i tested against Yakalir. We don't timeout that in the arena. But if i play against him in the IDE, one of us will always timeout at the first frame. Always. No matter the seend.

But if i play against reCurse, none of us timeout.

@Magus, if you have a given seed and you play first against Yakalir and you timeout and then with the same seed you play first against reCurse and you do not timeout then this would demonstrate a defect on our platform as the other player must not be able to impact your player in any way. So if you come across such a scenario, please let me know.


#20

If i remember i'll try later


#21

Quick test I did this evening: if I play against Magus, then I will face a timeout on his bot. If I swap positions, then it's my bot that is generating a timeout.
So it seems it is always second player that is impacted.

Checking my code, I can confirm that during first turn my bot is assessing he has 1 second left to answer then 100ms for next ones. Others bots facing the same problem could have implemented the same logic.
Then, if I reduce my first turn timeout to 95ms, then I do not face any more timeout...

Could the upgrade have affected that assumption in any way?