The interface receives messages for events that are no longer useful, it displays errors, etc.

Right now the UI freezes.

Only call getCourses at meteor startup, an maybe every few minutes.

At SO, there is a problem with the vmchecker login form in Internet Explorer browser:

You enter the login data (username and password), hit Enter or click on Login and nothing happens - the fields get empty and the user remains on the login page.

The horizontal scroll in the popup window decreases usability.
Two proposed solutions to solve this:

1. Resize the popup window so it's width fits the text inside it. This could create problems with very long lines though.
2. Automatically insert a line break in lines larger than X characters and resize the window to fit X characters in width. X could be a classic value like 80.

See the email thread [0]

[0] http://lists.rosedu.org/pipermail/vmchecker-dev/2009-April/000623.html

update_db.py reads grade from the first line of results/job_results.
however it should add different points found in the file, apply penalty and save the mark

results/job_results might look like:

+0.1 implementare
-0.2 memory leak at line 666
-0.3 possible deadlock

update_db.py should add 10 + 0.1 - 0.2 - 0.3 then substract the penalty, say 0.5, to get the final grade 8.9

Docker (https://www.docker.com/) would be a great backend for most use cases

Sometimes, it's necessary to display some sections of grade results first, because of ergonomy reasons. Please design a generic mechanism for setting a custom display order for these sections.

There are two issues stemming from users selecting course after logging in:

• students mess up and send the homework in the wrong course
• messed up configs for one course can bring down other ones (e.g. the auth_config.json file)

We need the student's name when rendering the global results table, in order to display the name nicely. We also need the student's id when the authentication is performed.

http://click.pocoo.org/ could possibly be a good alternative to optparse

There should be a way to check that an assignment has all required fields in .vmcheckerrc eg: course, deadline, timedelta, etc and that tests exists.

We want to be able to:

• see list with history of uploaded assignments
• download from vmchecker last uploaded assignment

It should work on Linux, but it doesn't. Log file is empty.

Update user interface to only allow uploading of ASCII text files. If user uploads files that use another encoding, the upload process should fail.

Some server methods have a username parameter.
Remove that in favor of this.userId

https://github.com/rosedu/vmchecker/blob/master/meteor-ui/vmchecker/server/server_save_file.js#L6
https://github.com/rosedu/vmchecker/blob/master/meteor-ui/vmchecker/vmchecker.js#L390
etc...

Visualization of the testing queue. We have define a new service and to find proper place for displaying the queue.

Currently, you need to click the dropdown, for the first course to be selected.
This should be done automatically.

In the current architecture, each assignment is associated with a virtual machine and a checker. There are situations where this modes doesn't match the needs of the assignment.

For example we want a mechanism where a submitted homework needs to run on a virtual machine and then run the same on different virtual machine. Or another case where the virtual machine is not on the checker machine and the vm should first be copied on the tester.

I am proposing a rather large feature request: a sort of a script before the checker script for the setup before the run. A pseudoscript for this could be something like:

upload assesment on vm1
result1 = vm1.runchecker()
upload assesment on vm2
result2 = vm2.runchecker()
return (result1+result2)/2

Or:

get vm1 from ssh://admin@server/home/$student/vm.zip
unzip vm1.zip
upload assesment on vm1
result = vm1.runchekcer()
return result

A usecase I'm thinking:
Have another account, say [email protected] which students can use to submit their homeworks without using the the upload interface. The only problem would be the invoking of callback script which can contain almost anything.

At SO & SO2 courses it was suggested to allow students to resubmit their homeworks. This has the potential of a DOS attack.

If a user logs in writing his username with capital letters he will be in a new blank account.
(Ex: firstname.lastname vs FIRSTNAME.LASTNAME).

His name will be a duplicate in the results page too.

Screenshot :
http://imageshack.us/photo/my-images/32/pozar.png/

Currently, when you select a file, it is silently uploaded.
We need a way to display that it was successfully uploaded, or that there was an error

See title

queue_manager doesn't check that there can be more submissions of the same assignment from the same user in the queue. so if somebody sends 5 archives in a very short time (for example by double clicking or fixing a missed bug) all 5 homeworks will be evaluated.

I've come across an output file (run-stdout.vmr) of 82MB in size. This should be taken into account. It messes:

• the storage
• the database used (currently Git)
• the interface (it wouldn't be able to list such a large file)

Output should be limited for each assignment to a given threshold.

Most of the increase in output size is due to left out print statements and infinite loops in the assignment code.

The vmchecker init script (placed in /etc/init.d/vmchecker) is not LSB compliant. When upgrading insserv, the following error ocurs:

error: Unable to migrate to dependency based boot sequencing.
error: Problems detected: insserv: warning: script 'vmchecker' missing LSB tags and overrides,  , package klogd
removed but not purged, package sysklogd removed but not purged

See http://wiki.debian.org/LSBInitScripts/DependencyBasedBoot for
more information about dependency based boot sequencing. To
reattempt the migration process run 'dpkg-reconfigure sysv-rc'.

Testing queue displays the following message (for all courses):

"Coada temelor ce urmează să fie testate

{
"errorTrace": "Traceback (most recent call last):\n File "/usr/lib/python2.5/site-packages/vmchecker/websutil.py", line 239, in get_test_queue_contents\n look_for_keys=False)\n File "/usr/lib/pymodules/python2.5/paramiko/client.py", line 291, in connect\n sock.connect(addr)\n File "", line 1, in connect\nerror: (113, 'No route to host')\n"
}"

There should be a specialized link name (such as "c") for plagiated assignments. "ok" means submission OK, a number means the assigment grade and "c" would mean plagiated assignment.

When deadline for an assignment passes, the default selection for an assigment should be updated to the new assignment.

This would helpfully prevent situations when students upload a new assignment archive over an old assigment.

If there are no tests defined for the assignment (a zip archive in the tests folder) the upload operation displays an error message. However, the archive is uploaded and saved in the repo.

Explore how team assignments could work from vmchecker's perspective:

• how to create a team (all members must agree);
• once a team is created, notify someone (TA?);
• who uploads an assignment: all members or just one;
• if only one team member uploads assigment, make it visible for the other members as well (both in results and in TAs view).

Currently the first page form does not allow my browser to remember my username and password used on vmchecker.

I think I found some useful information about it here [1] under "How to remember login"

http://code.google.com/p/google-web-toolkit-incubator/wiki/LoginSecurityFAQ

Define a list of files that must exist in an uploaded archive (e.g Makefile , there is no reason to send a homework to checker if there is no way to compile it).
Also , think about a black list of files ( e.g , object files) that shouldn't be found in an archive.

There is only a summary with no. of errors and no. of warnings, but there is no detailed list of warnings. Also, there seem to be about 1-2 warnings that are present in all uploaded homeworks.

vmchecker-resubmit doesn't crash if the --user parameter is incorrect

Some bash completion scripts would be useful for assignment & user auto filling.

Logging in with mixed lowercase or uppercase combination of the same username and uploading a homework generates two different entries.

Reproduction:

• log in with username and submit a homework
• log in with uSername and submit the same homework
• check the grade list

The submission appears twice, and seems to be stored as the users are completely different.

It's possible to run two queue_managers simultaneously that listen on the same directory. This screws things up and can hang the virtual machines. A locking mechanism should be implemented so that no two queue_managers listen on the same directory.

Trailing spaces are not removed from the login username. Therefore, username, [spaces]username and [spaces]username[spaces] are considered three different users and generate three different entries.

This bug seems very similar to issue #20.

When selecting an assignment, we need to display the current status:
None, Uploaded, Tested, Graded, etc

Provide an URL for each assigment. That is, map an URL to the results page of an assignment.

My suggestion is to use a hierarchical URL, such as http://elf.cs.pub.ro/$class/$assig_id/$student, where $class is the class name, $assig_id is the assignment id and $student is the name/ID of the student.

From SD course:

xxx.PADUREANU ok
xxx.Padureanu ok
xxx.pAdureanu ok
xxx.paDureanu ok
xxx.padUreanu ok

Usually during holiday no penalties should be applied.

Let's take for example SO course and the Easter holiday: April 20th - april 27th. The deadline was on 17th of April.
A homework submitted:

• before 17th should take full points
• on 18th will have 0.25 points of penalty
• on 19th will have 0.50 points of penalty
• during the holiday will have 0.50 points of penalty (penalty doesn't increase)
• on 27th will have 0.75 points of penalty

This leads to very slow responses when the user uploads something.

This is because we don't update the files when a new evaluation result comes but when we update the UI.
This leads to very slow responses and messes up the message queue in the UI.