Tests of internal mechanisms

Submissions # 5225109 and # 5225113 (for spoj.pl/problems/ADV04L/), the time is different (The difference is 0.01, which migh not be much, i know, but a bit more signifcant errors had happened, i just dont remember right now.)

Judge for spoj.pl/problems/PRIC/1 doesn't work correctly.
Former AC submissions produce WA now, even a text submission of the sample gives WA.

I get (see e.g. submission spoj.pl/error/53352501) this weird compilation error.

{standard input}: Assembler messages:
{standard input}:3: Fatal error: can't write /home/Jw2bFh/cc9VC8Sf.o: No space left on device
{standard input}:3: Fatal error: can't close /home/Jw2bFh/cc9VC8Sf.o: No space left on device

The source file is 3k and the code generated on my machine is 13k in size, so nothing that could really eat up much of disk space.
Moreover, I see a lot of other people failing C++ submissions with compilation errors tonight (though a few get past the compilation hurdle)

Different times for same code. SubmissionID: 5346787 and 5346790. Problem > spoj.pl/problems/CANDYSTN/

Judge seems to have some problems: There are several 0k AC submissions actually.

Solution AC with 0k memory. Submission ID: 5348366. Problem > spoj.pl/problems/GCPC11F/

There are more such 0k solutions - not only AC, but also WA.
Moreover, the problem concerns all contests.

I copy pasted the official test code at: spoj.pl/forum/viewtopic.php?f=31&t=26 and got NZEC runtime error, same for any other problem I tried to solve using Prolog, even if I don't output anything.

What is wrong ?

There is definitely something wrong with the actual Python 2.5 implementation.
It seems that there are different implementations used on different servers.
Please have a look at my submissions 5357523 and 5357524. The code I submitted is exactly the same, but runtime and memory usage differ significantly. Standard runtime for a Python program that does nothing more than starting the interpreter used to be 0.03 s for Python 2.5 and memory usage used to be 3.7 MB. This is congruent to my submission 5357523. But submission 5357524 takes 0.18 s and 4.4 MB for exactly the same code.

Same problem with Python 3 installation. Two submissions of the same code:
Submission 5357623: AC in 0.91 s with 5.6 MB mem
Submission 5357624: AC in 2.64 s with 6.6 MB mem

Another one: The Python 3.x installation lacks the random module. See my submissions
5357566 (AC) and 5357567 (NZEC) - same code, but random module imported (and not used!).

I can absolutely confirm this. Something must have happend. The results do significantly differ with identical code.

I don't know what this statement refers to, but the Python problems still exist.
- Python 3.1 has no random module available
- Runtimes for both Python 2.5 and Python 3.1 seem to be stable, now, but to the worse:
Python 2.5 needs about 0.15 s and 4.4 MB mem just for startup compared to 0.02-0.03 s and 3.7 MB mem which was typical before.
I don't understand why the former Python 2.5 installation that worked fine, cannot be left as it was?