dalejbarr / assessr Goto Github PK

feedback_all() should use safely() to avoid this behavior

loading everything into memory can be a problem with large classes, perhaps give user a warning? also, should save as you go rather than only after clicking Done

list seems unnecessary. Simplify code

compile_key() should give clearer error messages when solution/assessment code chunk names don't match

from emily nordmann:
fetch() is now giving:

Error in curl::curl_fetch_memory(url, handle = handle) : Failure when receiving data from the peer

Would be handy sometimes if the assignments were ordered by code similarity in the browse stage - or have the option to change the sorting if other alternatives come up?

also check NaN

currently creates a weird error if one of the values is logical

Instead of searching for and commenting out dangerous functions (install.packages()), replace them, e.g.:

install.packages <- function(formals("install.packages")) {
   assessr::add_feedback("* use of this function is forbidden")
}

Moodle only accepts zip files < 100MB, so something that would calculate file size and break into subparts for upload

If a student types:

#---------------------

This is treated as a chunk delimiter (because of knitr::purl())

useful for inspecting ggplots

Do we need this? Would make the assessment more robust as collaboration still encouraged but need to work on their own data. Would be good to chat about this with team.

If you put a colon in the title of the feedback report you will get this (frustrating) error

 Error in yaml::yaml.load(string, ...) :
   Scanner error: mapping values are not allowed in this context at line 1, column 24

I think it has something to do with the remove_comments option

from Phil:
on lab 5, task 2, the same dimensions call as the solution dat2, doesnt quite work. There is something different about dat2 <- select(dat, Intellect_Rating, Impression_Rating, Hire_Rating, ï..CONDITION) and dat2 <- select(dat, Intellect_Rating, Impression_Rating, Hire_Rating, CONDITION)`

• disallow unnamed chunks
• disallow duplicate chunk names
• make sure the solution file and stub file will knit

Need to re-think the workflow. Currently it is tricky to do a mass assessment and then re-do specific assignments (e.g., late submissions, or specific problems)

this causes problems when IDs start with 0

Important when having different task orders for different submissions

if a student puts question marks in the code chunk:

read_csv(?ahi-cesd.csv?)

the browser hangs

for instance install.packages() will run if in a block that is not assessed