You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Could be useful for example if the isolated program assumes that the input is contained in a specially named file, say, input.txt
I think this feature would provide a good solution for Checkers #32 – I doubt adding the "checker" step would be generic enough for all users; I think a better approach is to keep the API on the low level let the users orchestrate checking by themselves. This could be easily achieved if the checker can be fed multiple files, e.g. input.txt, output.txt, answer.txt (aka expected output). The caller can then analyze the text output generated by the checker, in, say, verdict.txt).
Some programs span multiple source files. Automatic systems could compose the contestant's code with preset sources with the test data already in place, for example, so that instead of implementing reading the input and writing the output the contestant can focus on implementing a function that solves the problem. Composing multiple source files into one is very hard to do right generically in most programming languages.
I believe all these issues can be solved by simply provifing the multiple source/input/output files feature (or even better, arbitrary isolated directory layout). Thoughts?
The text was updated successfully, but these errors were encountered:
sorry for the late response on this topic. As I am more and more interested in implementing this feature I would like to start discussion about it.
Given the current API, how would you suggest that the API call needs to look like to support this feature of sending multiple files that will be located in the sandbox?
Do you think it is worth designing the API such that a virtual environment can be reused between submissions?
Personally, I would love to see a more low-level interface to isolate, that exposes its features through API calls. I.e. creating sandboxed environments, uploading/downloading files, running programs. That would be a big change though, so I'm not sure that it is feasible.
Use cases that I see:
input.txt
input.txt
,output.txt
,answer.txt
(aka expected output). The caller can then analyze the text output generated by the checker, in, say,verdict.txt
).I believe all these issues can be solved by simply provifing the multiple source/input/output files feature (or even better, arbitrary isolated directory layout). Thoughts?
The text was updated successfully, but these errors were encountered: