MOP: Score generator replaced by exporting of points

[eval.git] / TODO

diff --git a/TODO b/TODO
index f31fbe38cd8b01f9d112898b76f99d47aa8c1fc7..698a6ea58eaa1a54a4bf075bb63d2772d80f941b 100644 (file)
--- a/TODO
+++ b/TODO
@@ -1,27 +1,39 @@
 Evaluator
 ~~~~~~~~~
-score: better formatting of the score table
 interactive tasks: solve deadlocks?
 interactive tasks: logging of messages
 terminology: problem -> task ?
-writing of verdicts to stdout (include timing and check on interactive tasks, including judge errors)
-box: memory usage statistics (need to search for VMPeak in /proc/$PID/status, but it disappears too early)
+
+Judges
+~~~~~~
+test cases for judge-shuff
 
 Environment
 ~~~~~~~~~~~
-fpc: add switch -gl (runtime errors write line#)
+fpc: add switch -gl (runtime errors write line#), but avoid that when inside the sandbox
 
 Installer:
-- mo-install: reduce what gets copied to mo-eval home
 - update group permissions
 - quotas
 
 New submitter:
-- Checking of contest time (and per-contestant exceptions)
-- contest: override failed check
-- contest: local history
 - contest: task status cache
 
 Various ideas:
-scores dependent on time (like PL olympiad)
-box: konfigurovatelny pristup k timerum a /proc/self/stat (pripadne /proc/self/fd)
+- scores dependent on time (like PL olympiad)
+
+TomGav
+~~~~~~
+recursive try-ln for dir input
+clean and well-defined expansion (beware of per-test re-expansion)
+configurable names of in/out files (independent of $PROBLEM)
+paranoidly check file mode/owner before running the sandbox
+
+Isolate
+~~~~~~~
+Test: ptrace self
+Test: SIGSTOP
+Test: ping-pong timing attacks
+Test: big static memory
+Examine the use of taskstats for measuring memory
+Switch license to GPL2/GPL3