Time_limit type (float | int ) & >0

[thorehusfeldt]

The 2025-09 draft has

optional float > 0, see Problem timing

as well as

The value of time_limit is an integer or floating-point problem time limit in seconds.

I think the latter is right.

[niemela]

Strongly agree!

[mzuenni]

Depending on how you read the spec any float in any files is allowed to be an int?

All floating-point numbers (in any files, including in the problem package or submission output, that are parsed by contest system tools or the default output validator), must be given in decimal and may use scientific notation. More specifically, floating-point numbers must satisfy the following grammar, which accepts formatted floating-point output from most major programming languages:

sign         [+-]
digit        [0123456789]
expIndicator [Ee]
significand  ( {digit}* "." {digit}+ | {digit}+ "." | {digit}+ )
exponent     {expIndicator} {sign}? {digit}+
float        {sign}? {significand} {exponent}?

• The problem.yaml is a file, included in the problem package, that is parsed by the contest system.
• any int matches {sign}?{digit}+ and thus is a float without {exponent}?

I think this is pretty convenient because 1 is shorter than writing 1.0 and it's clear what this means.

However, currently this is somewhat "hidden" (or even unintended?) in the Spec. Maybe we want to replace all places where we specify float with int | float?