[LON-CAPA-users] changing part name
Guy Albertelli II
Wed, 16 Mar 2005 10:43:51 -0500 (EST)
> So republishing should mix up how the CODEs for the anonymous match with
> the students answers? I don't really understand how that works. Is it
> just that the CODE is somehow used in the random number generation?
Everything releated to CODEs will work fine. There is no relationship
between CODEs and usernames. (CAPA was braindead in the way it did
this, blame a younger more inexperienced Guy for the silly way it did
In Lon-CAPA it all works perfectly. :-)
To explain how it works, normally the random numbers are based on the
username, the domain, the internal coruse ID, and some aspects of the
problem (the map url, the problem url, the id number of the problem
in the map, and in some cases the part and response ID numbers)
When you print 'Anonymously' with CODEs, the CODE replaces the
username when building the randomseed. And the exam is printed out,
there is no relationship between users and CODEs. And the CODE is a
source for the randomseed, not a result of the random function.
When grading the system reads the exam, discovers the CODE for the
exam that was used, discovers what user took the exam and at that
point time says grade this users exam with this CODE.
So feel free to republish it won't muck the CODEs.
> > > I made the mistake of having an "=" symbol in a part name in a problem.
> > I'll make the assumption that you mean 'part ID' and not 'part name'
> > > This is making trouble for evaluating grades in the spreadsheet (the
> > > spreadsheet doesn't recognize any of the parameters). The problem was
> > > used in an anonymously generated, scantron-graded exam (all the grades
> > > have already been uploaded).
> > Are there any error messages in /var/log/httpd/error_log?
> > > I changed the part names in construction space, but am afraid to republish
> > > because I am not sure what it will do with the uploaded scantron grades.
> > > Even if it changes the randomized numbers in the problems, at this point
> > > it doesn't matter (the question is closed), but does anyone know if this
> > > will mix up whether or not the student received credit.
> > Some details:
> > 1) curently assigned credit is assigned to 'part ID's (Meaning they
> > get a specfic part ID correct or incorrect
> > - thus by changing the part id and republishing, the old part ID is
> > no longer in the problem and the new part ID is not correct, which
> > appears to user as if they 'lost' credit, the data about the
> > correctness on the now missing part id is still stored just not
> > shown
> > - the randomization of &random() is not based on 'part ID's
> > - the randomization of <option/radiobutton/ranks/match/image/response>
> > _IS_ based (partly) on 'part ID's
> > So if it is a <numericalresponse> in question the process of
> > 1) change the part ID
> > 2) republish
> > 3) regrade the scantron form for all students
> > Should result in every student getting a correct grade assigned.
> > > Also, are there set guidelines for what can and cannot be used in a part
> > > name (special symbols, spaces, etc)?
> > I wasn't aware of anything not working. Off to plumb what = is screwing
> > up.
> LON-CAPA-users mailing list
firstname.lastname@example.org LON-CAPA Developer 0-7-2-9-