CVS log for loncom/homework/response.pm

Up to [LON-CAPA] / loncom / homework

Request diff between arbitrary revisions

Keyword substitution: kv
Default branch: MAIN

- Removing training white space and one blank line. No code changes.

- For 2.11
  Backport 1.250

- Bug 6995
  Space (ascii 32) is not supported as a delimiter after \verb for TeX Live
  2020 and later. * (ascii 42) should also not be used as a delimiter.

- For 2.11
  Backport 1.249

- Bug 6680
  Use value(s) from any responseparam tags present in the template itself
  when grading submissions to example templates.

- For 2.11
  Backport 1.248

- If questiontype in effect is one of the four survey types, and award on submission
  for a customresponse item is 'ASSIGNED_SCORE', change that to one of:
  SUBMITTED, SUBMITTED_CREDIT, ANONYMOUS or ANONYMOUS_CREDIT, as appropriate.

- For 2.11.2 (modified)
  Include changes in rev. 1.247

- For 2.11
  Backport 1.247

- Reinstate "Submit all" functionality in multi-part problems in composite page.

- For 2.11
  Backport 1.246

- Support retrypartial parameter in &check_status().

- <partorder> tag for a library file will include ids of any part tags.
- library file is only loaded and inspected to check for parts if <partorder>
  is absent.

- Bug 6802: support Math Previewer for all textboxes in formularesponse
  with multiple answers.
  - Move &edit_mathresponse_button() from response.pm to inputtags.pm
  - Use HWVAL_$input_id instead of HWVAL_$partid_$responseid as unique
    id attribute for each <input> tag created by a LON-CAPA <textline />.

- To support old browser versions use DragMath pop-up instead of inline
  math preview (javascript); determination based on user agent.

- Previewing math expressions by a student (mathresponse and formularesponse).
  - Course/Community defaults includes a domain configuration to specify use
    of the LON-CAPA Math previewer (can be overridden in Course Settings --
    "Display of Resources" -- by CC).

- Modify how information from user agent is used to determine whether pop-up
  dragmath applet (java) or inline math preview (javascript) should be
  offered. Work in progress.

- Bug 6476
  - Store rawrndseed for parts with all responsetypes.
  - Eliminate storage of rndseed, since rawrndseed is the item needed to
    reconstruct the randomization for a problem in Authoring Space.
  - If questiontype is randomizetry rawrndseed and rndseed are the same,
    so use rawrndseed from the submission history, where it exists.
  - Support use of randomizetry for script blocks outside a part block in
    problems with a single <part></part> tag.
  - Support use of randomizetry for script blocks inside part blocks.
Note: randomizetry does not apply to randomization in script blocks
outside part blocks for problems with multiple parts, because the
number of tries is not uniquely defined.

set implicit operators for mathresponse preview

- Changes in args passed to &edit_mathresponse_button() to support use of
  id attributes by LON-CAPA Math editor.
- DragMath implementation now also uses id attributes.
- Support math editor in formularesponse and mathresponse items in
  composite pages (.page).

added new LON-CAPA math editor for mathresponse and formularesponse to replace the DragMath applet

- due date checking in &check_status() needs to support time limits.

Internationalization: Added missing &mt() calls

- Bug 6685
  - Sign DragMath .jar files with InCommon code-signing cert from MSU
    (expires Dec 4, 2014).
  - Update to DragMath 0.7.9
  - Installation mechanism now removes files from older DragMath version,
    and follows similar mechanism to one used for MathJax etc.
  - xhtml for EditMathPopup.html and MaximaPopup.html
  - codebase changed from classes to applet (DragMath change: 0.72 to 0.7.9).
    - EditMathPopup.html and MaximaPopup.html now installed in:
      /home/httpd/html/adm/dragmath (paths in response.pm and
      lonhtmlcommon.pm updated accordingly.

- Appropriate award if question type is for a survey or anonymous survey
  for customresponse and mathresponse.

- For perl 5.16 use of deprecated defined(@array) generates warnings.
  - eliminate use.

- For perl 5.16 use of deprecated defined(@array) generates warnings.
  - eliminate use.

- Bug 2802.
  - Separate decision on leniency in grading from logic on whether additional
    new submission is possible for partially correct problem parts.
  - new parameter: retrypartial set to "yes" if part is to be answerable
    when partially correct.
  - custompartial.problem template incorporates retrypartial parameter in
    place of lenient parameter.
  - points awarded > 1 no longer considered 'partially correct'.
  - &grading_is_nonlenient() moved from inputtags.pm back to optionresponse.pm

- Backport 1.229.

Displaying previous tries for customresponse.
  - Actually use &format_prior_response_custom().
  - Check if submission is an array (e.g., custompartial.problem template).

- Bug 2802.
  - optionresponse::is_nonlenient() moved to inputtags::grading_is_nonlenient()
    to facilitate re-use.
  - where solved is "correct", awarded value < 1 permits new submission,
    only if lenient grading applies.
  - add lenient parameter to custompartial.problem template.

Using bubblesheets to record grades for handgraded exams
  -- choose what filled last bubble represents (either a zero or max,
     i.e., 5 for 5 bubbles per line, 10 for 10 bubbles per line etc.)
grades.pm
  - New routine: &hand_bubble_option() determines if sequence to be graded
    contains response types which are handgraded, and offers radio buttons to
    choose score for last bubble.
response.pm
  - &repetition() routine can be called in scalar context of in array context.
    scalar: returns number of lines to encode weight
    array: returns (number of lines to encode weight, bubbles/line).
chemresponse.pm imageresponse.pm caparesponse/caparesponse.pm
  - &response::repetition() now explicitly called in scalar context.
Work in progress.
  (Corresponding option needed in lonprintout.pm to set
   correct number of bubble rows for handgraded questions with
   weight >= bubbles per row).

- Backport 1.225.

- Additional (eighteenth field) BubblesPerRow in Bubblesheet Format file.
  - specify bubbles available for each question row (default is 10).
- Support case where question (or subquestion for optionresponse,
    matchresponse, or rankresponse) requires more bubbles than in one line.
  - Students prompted to bubble once in multiple lines.
  - Numbering recognizes multiple lines per question.
  - Also supports: imageresponse, stringresponse, formularesponse where
    students leave link(s) blank if weight assigned is greater than
    bubbles per line -- multiple lines also supported in this case.

For example:
- 15-bubble MSU Art History exams now supported on standard 10-bubble
    bubblesheets used at MSU.

- Stop false positives.

Interestingly, one needs the pre-processed random seed to reconstruct
the problem in construction space. Now stored as rawrndseed.

Bug #6476: reversing lonnet.pm 1.1109 to 1.1108
Needed to catch post-processing of random seed.

Saving my work: lenient grading. Still problems with saving input after
first try.

- New Question Type - randomizetry
  - Additional args to &setrandomnumber() -- $target and $rndseed
  - Additional arg to &setup_prior_tries_hash() -- $questiontype.
     If $questiontype is randomizetry:
       When setting up prior tries display deternine options array
       (for rankresponse, optionresponse and matchresponse) from part's
       foilorder key in Apache::lonhomework::history hash

Need more room for good answers

- Bug 5531.
  - Detect where student resubmits previous answer for case sensitive
    stringreponse item and award changes to EXACT_ANS.
    Detect for case insensitive stringresponse when previous award changes
    to EXACT_ANS for submissions whch only differ by case.
    - logs to lonnet.log and sends e-mail to recipients of "errormail" for
      course domain.
  - For cs and ci types if award is INCORRECT use perl to compare
    response with answer, and log if disagreement with result from
    strcasecmp() or strcmp() to capa_check_answer() in capaCommon.

- Typo in pod.

- Bug 6119 Anonymous Surveys.
  - additional survey types: anonsurvey, anonsurvey_cred, survey_cred.

Bug #2802: partial credit multiple answers

XHTML:
- XHTML conform <img> tag closure
- Added some missing alt attributes to <img>

Support for R CAS
- needs testing
- better error handling
- correct blacklist
- correct list of allowed libraries

XHTML conform javascript start tag attributes

- Protect against the effects of using partIDs and responseIDs which contain characters with special meanings in perl regexps.

- Verification of Scantron grading for problems containing randomlists
  - &scantron_partids_tograde()
    - changed to use &get_analyze() to retrieve partIDs/responseIDs
    - subroutine moved to be within scope of %analyze_cache.
    - needs to be called for each user (randomlist -> possibly different responseIDs for different users).

- Robustness against changes to lonxml::counter being made by multitasking instructors.
  - scantron_questnum_start.$part.$id form element passed in ssi call for scantron grading.
    - where available, this value is used within response::getresponse() instead of lonxml::counter to determine starting column for scantron line.

- Second pass (if optional inbuilt verification enabled during scantron grading).
  - Added missing call to repeat &grade_student_bubbles() when anomaly is detected during first pass.

****
Work-in-progress - possible performance hit - perhaps check if resource includes <randomlist></randomlist> before calling &scantron_partids_tograde() for each user?

- Restore pre-1.207 functionality to &mandatory_part_meta().
- Eliminate previously commented out code (commented out in rev 1.34 - 8/2001)     - moved to POD in 1.207.
- Eliminate unbalanced 'back' from POD.
- Remove args from subroutine titles in POD items so perldoc underlines them.
- Move POD for &mandatory_part_meta() so it precedes subroutine.

- Backport 1.209.

Detect if there were timeouts or other internal errors during grading.

Do not charge a try, allow student to enter same answer again.

POD documentation changes

POD changes

Localization, wording, nobreaks for Colorful Editor
- Added missing &mt() calls at several places (more to be done)
- Same wording "Delete?" for all Delete options
- Same wording style, e.g. added ":" after text and before optionlist
- Added nobreak style for some elements which need to stay together, e.g. help_open_topic in loncommon.pm
- Added/updated some related German translations

DragMath for mathresponse and formularesponse.
- As the default is to provide the DragMath Editor - a parameter: turnoffeditor (default "no") titled "Hide DragMath Editor" is more consistent with usage for existing parameters.

Past tense of "occur" is "occurred" (rr!)
Also added some missing &mt() calls and corrected some other little typos

Added ressource parameter to turn DragMath-Editor on/off.

Moved the pencilicon behind the textfield for mathresponse.
Fixed bugzilla #5772: Implemented turnoneditor as an attribute for formular- and mathresponse.

Removed helplink from DragMath icon.

added title attribute to pencil-icon.

Corrected pencilbutton behaviour when question already answered.
Modified pencilbutton style.

Modifications to pencil.

*** empty log message ***

Fix icon path

added stift.gif

added helper

Some documentation

Bug #5738: be able to load libraries into Maxima

Dragmath now also in mathresponse

Debugging log removed

Bug #4501: answer display too wide for columns

Bug #4501: printing answers runs all over the colums. Now takes up more
vertical space, though

When the target is 'analyze', include the responsetype of the current item in the lonhomework::analyze hash, for use later on, e.g., for scantron grading.

Bug 5584

- xhtml

BZ4074 - Tested for multiquestion exams.

BZ4074 - Complete testing of single question exams for bubble sheets
in number and positional format.. removed log spew where needed.

BZ4074 - I think the grades finally work full circle for at least a single
question exam.

 - BUG#5395, split feedback off into two options, no feedback at all, and no feedback on correct/incorrect (but allow unit/malformed answer feedback)

BZ 4074  Ensure the bubble lines chunk of the analysis hash gets
keyed by the concatenation of partid.resopnseid.

bz4074 - Use lonxml subs to maintain the number of bubble lines required
for responses to these types of problems.

- we want to save the exact bubbled letters

- documenting

- typo sigh

- correcting the reg expression to fill in all * needed

Two changes in &implicit_multiplication:
[1] (debateable)
Replaced \w by [A-Za-z0-9]. This assumes that in a student's answer a variable name never starts or ends with underscore. On the other hand this makes it possible to use underscore as an operator (e.g. a _ 3 for (a,3) is member of a relation. Can't think of a better example...)
[2] (required) Hitherto
$expression=~s/(\w)\s+(\w)/$1\*$2/gs
or
$expression=~s/([A-Za-z0-9])\s+([A-Za-z0-9])/$1\*$2/gs
transform
6 a b c d e f
into
6*a b*c d*e f
whereas
while($expression=~s/([A-Za-z0-9])\s+([A-Za-z0-9])/$1\*$2/gs){pos($expression)--}
will produce the desired output
6*a*b*c*d*e*f

- =pod commands need a blank line above and below them

Factor the multi-line bubble stuff out of radiobuttonresponse.pm and put
it into response.pm

- update docs on getresponse

- fix forgotten line in variable rename

Add my for $temp variable so that this can actually be loaded in
mod-perl without compilation errors.

- adding some docs to some internal routines

- lonxml takes care of all insertions now, and thus the tags that were attempting to handle it themselves could cause double insertions so elimnate external insertion calls

- add in mathresponse mathhint customhint (and their associated answers) for Edit mode insertion

typeo

Comment documentation for get_response_param

- 2622 for math, external, custom

- 2622, refactoring out some code, essay response also works now

- protext from the tag having special charecters in it

- implement some of the suggestions from bug#2622 about prior tries display

- BUG#5203, trying to be smarter about when the user presses enter (should now correctly submit the correct field)

<customhint>

<mathhint> functionality

Make formula fix work with <mathresponse>

More elegant

<mathresponse> appears to work, but needs lots of testing

Saving work for MAXIMA.

- expecting an array of values in answertxt now (for when there are multiple input fields)

- scripts must return a valid award string '1' and '0' aren'a valid award strings

Just saving my work on Maxima - currently stuck with Bug #5102

- few tailing whicuser conversions

- trying to preserve the fix for BUG#4250 but use the normal answer functions to do so

- adding ability for multiple answer for 1 response

- making custom response point scorable
- adding an answer display to customresponse (BUG#4938)

- moving the help to the 'correct' location

Document custom response

- check_status was not working correctly with part 0

- need to properly handle the no <Instance> case of Id picking

- updating the random seed to look in the correct location for dimension and instance ids

Model <mathresponse> after <customresponse>

Moving mathresponse

- sigh precendence

- rank/match/option response were not display the correct bubble line number
  in PGRD of exam resources

- switch id getting to use new mechansim for ids

- more of BUG#4608, was loggin survey submissions to the grading queue

- save the space by no indenting the answer

- better handling of the header bit
- allow for option increment to the counter (for multiple answer lines per part)

- tex mode answer printing puts the wole answer bit in a single line now
- printing in exam mode puts the counter value in the answer mode print

- only add to queue if we are in a course and working with a published ymbed resource

- start of de appendixing grading queue

- algorithmic -> custom

- <algorithmicresponse> BUG#4451

- removing some debug crap
- soime styl work
- allow a pushrandomnumber to specfiy to only use the primary id number (<part> id or <dimension> i) to modify the random number generator

- 64bit5 more randomish looking, no longer trying to munge on our own, using Digest::MD5

- document the essayresponse upload scoring
- fix BUG#4385 (printing extra lines for scoring)

- bug#4279 | could show up as a valid character in an answer, now we search for a nonexistant char in the string and use taht one

- Fix BUG# 4263

BUG#4185, the submit buttons on .page broke due to the part by part submission change

- need to pick a defualt weight for CSTR (BUG#4089)

- ENV -> env

- making bridge tasks randomize whihc instance in a dimension is shown

Correct error in figuring out privilege to show all foils.

- BUG# 2331, homework can be done in 'sequential' mode now, Only show part 2 once part 1 is correct, or unanswerable (tries exhausted)
- added new function &check_status($partid) to check the current status of a part

Decide to do print all foils by removing the requirement for a special
printing env variable (Guy's suggestion), and instead, remove the requirment
of a symb being absent in adv. roles.

Add latent support for printing all foils of e.g. option response problems:
this is being done via the addition of print.showallfoils to the
or check in sub showallfoils

- returnning the end_responseparameter correctly

- Part of #636, Submit Answer now only submits for a single part. The logic for detecting a submission has been broken out to a subroutine if future tweaks are needed.

- make response:setrandomseed more randomish
- fix BUG#3760, numerical response exam mode borked when submitting response since I changed it to start using prettyprint, &make_numerical_bubbles now submits back typeset and un typset bvalue

- somtimes the response id is the only source of 'randomness', BUG#3627

- 64bit4 and numval3, numval2 could generate values greater than 2^31-1 in somecases (long symbs being a good one), which would collapse to -1, thus making the symb a source of no random seed data

- forward port 64bit rndseed changes

- 64bit fixes

- Bug#1 <or> handles survey mode

- previous submission handling now works for survey mode, in survey mode all previous submissions can be picked again, but we throw away a submission if the last submission is the same as the last previous submission

check_if_computed: do not assume we have a hash - was causing ISE on
stringresponse

- backwards

- die rankresponse stupid bug thing, Die Die!
- _really_ _really_ fix the bug where rankresponse answers were shifted by one

- my fix for scantron mode broke online mode, this should fix both

- slight speed up for CSTR space
- BUG#3431, turn on 'Show All Foils' when advanced users are browsing

- BUG# 3336, was waring incorrectly about not finding a <part> around a <response>

- BUG#3149, need to stroe the order of responses, so we can show them in the right order, all user of lonnavmaps are good to go now.

- fixes for randnumber generation
   - due to a error in coding the 64bit and 64bit2 rnd seeds were throwing away the second half of the seed, we now preserve this functionality and make it explicit
   - additionally 64bit3 now uses all 64bits of the seed
   - also added new function pushrandomnumber poprandomnumber to be called instead of setrandomnumber before we were crunching the Safe space random number generator state, as this function is used bu thing the <optionresponse> and <radoibuttonresponse> this meant changes in the internal randomization code could screw up later Safe space random calls the old method is preserved for blank 64bit and 64bit2 randomiztion env.

- was comparing to submissiongrading too, which is silly

- trying to be UTF8 transparent

- Start on BUG#2730, annotate in the analyze data structure whether an answer is computed or not.

- make hinttries work beter in CSTR

BUG#2758, $var in <responseparam> worked but didn't show that they worked in Answer mode

- added &mt() but not the use for it

- BUG#2340, throw errors on problems that are like
<problem>

<numericalresponse></numericalresponse>

<part id="14">
<numericalresponse></numericalresponse>
</part>

</problem>

- if we dont get a valid weight don't ISE, default to 1

- fixes <matchresponse> when used in scantorn exam
- <rankresponse> should be better behaved too. but I need to do some testing still

- BUG#2388, <responseparam> witha default of "0" was ignored

- Secondary randomizers (foil order etc) wasn't using proper method to get seed when playing with public versions of problems (BUG#2348)

- fixing up analyze mode for radiobutton/option/rank/match/image to all use the smae code

- new metadata info <partorder></partorder> contains a , seperted list of parts in order.

- backport changes to make problemstatus work better

- added &Apache::response::show_answer() to better check for if problemstatus inhibiting should be occuring or not

Bug 2061 is fixed. The content of the bug - Printing with Answers Before Publishing Broken.

- BUG#1676, all pacakge.tab parameter definitions are put into a drop down box to make it easy to add a known parameter type.

- Fixes BUG#1940, for new style rnd seeded problems with long textual part or response ids you could cgenerate a number larger than 2^32-1 which random_set_seed doesn't like. Now does a use integer in that section of code. Won't modify the results for normal problems.

Bug 1848 is fixed. Now it's possible to print answers in the same way as they appear in the problem body.

- BUG#1872, now more consistant

- missed commiting this on the last commit
- homework should be less dysfunctional

- consolidates the whichfoil functions and fixes BUG#1439

- propagating print_answer -> answer_output_mode shift

Changes which do not allow show up LaTeX chunks on the screen for the construction space.

Changes in answer_header, answer_part, and anser_footer to produce LaTeX output for the answer target neccessary to print problems with answers.

- response uses/understands new larger seed

- retabinate

- scantron grading of essay/formula/string works now
- new global Apache::lonxml::scantronmode can be used to check if we are in scantron mode or not

- break out the 'repetition' clculation
- essayresponse, start support of scantron

- consitancy for input fields

-also adding getresponse, as a better way for getting the current response

- Fixes most of BUG#1194, still need to work on editing

- Fixes BUG#965, multiple responses in the same part (AND style)
  now properly detect whether a situation is a previous submission or
  not, previously it would detect a situation in which the same individual
  response had appeared before (but not together) as a previously seen
  submission, now it properly only detects when a group of submissions
  is the same.

- Fixes BUG#991 (parameters were bleeding between responses)
- small formatting of answer mode change

- adding analyze to more tags in preperation for analyzing a problems answer distribution

- by Guy,

- <dataresponse> used to bacalled <datasubmission> for some silly reason I broke everythingby half renaming, now fully renamed.

- removed use of Perl's internal rand operator, now all randomiztaion depends upon the external Math::Random library

- nolonger wnat to do this limited loading of code, load all responses types at startup and register them

- implements BUG#238 - New radio button showallfoils, optionresponse shows
   all foils irrespective of the max parm and any conceptgroups, rbr and
   ir still need the functionality

*** empty log message ***

- %Apache::inputtags::params was being overwritten by hints, now start_hintresponse and end_hintresponse use Apache::inputtags::paramstack to save away values.

- picks correct rndseed for when viewing as someone else. (Fixes BUG#4)

- adding external response

changing "sub BEGIN" to "BEGIN"

Update to use edit::end_row() and edit::start_spanning_row() as a cleanup
for edit::select_arg and edit::select_or_text_arg.

- added <parameter>

- moved some of the previous cheicnking code into a function

- GPL headers

- added code so <responseparam>s will notify <*response> of their default parms (since &EXT() won;t be able too)

- <import> now puts info in .meta files abut what is being imported
- Apache::inputtags::import exists, list the <import> ids that we are in
- start_response, properly sets the Apache::inputtags::response id to the
   join() of Apache::inputtags::import and the id="" arg of the response
- <import> gets an id at publication time

- <radiobuttonresponse> display answers now

- added support for 'answer' parse to caparesponse.pm that works, <numericalresponse> and <stringresponse> both should work well.
- started adding <formularesponse> support

- randomseed is stable in construction space
- added helper functions so all answer printouts will look somewhat similar

- moved whichuser() from lonhomework to lonxml so I don't have to check for it's existance

- typos

- cstore/restore/rndseed check the environmetn to see if we are rendering
  for a different user or not (allowed calls need to be converted too)

- <dataresponse> displays some information about itself for target web

- renamed <datasubmission> to <dataresponse>
- improved <dataresponse> so that it is functional (kinda not to use parameter.html)

- typo
- fix so it only looks once per arg

- <numericalresponse> checks for all possible params

- support for the new parameter structure
- <numericalhint> properly sets up the Apache::inputtags::response so unique parameters are possible for hinting
- <numericalhint> gets a unique id

Does not create parameter default and display if not present.

Bug fix - stores were not stored.

-re asserting my indent style

beautify/optimize

- stabilized <randomlist> randomness

- moved the meta target run for numericalresponse

Write out part and numericalresponse packages to metadata

New routine to write metadata for packages of parameters

- detects previous submissions for <numericalresponse> and <stringresponse>

-implemented <stringresponse>
-fixed editor to remove trailing and leading spaces from user supplied values.
-updated documentation

- stopeed accepting " in tag arguments
- check if we need to hadnle insert for tags that have changed args

- added arg setting for <numericalresponse>

- added new arg $tagstack to all start_ or end_ routines

- modfiying to use the cleaner get_param calls

- redid random number generator for <i/rb/o/response> tags
     - should be unstable in preview mode and more random in normal mode

- </responseparam> needed to do something

- changing the description line on the tries data

- adding support for <essayresponse>

- <br> -> <br /> cleanup

- working on <imageresponse>

- numerichints should work properly now

- new response type <optionresponse>

- need to check for EXT sending back a con_lost

- nolonger cause ISE

- can now find and load radiobutton response types

- start on numeric_response
- starting on previous submissions detection

- adding the needed lectuure online tags
  not funtional currently but getting close

- moved funtionality around, all responses should call start_response, and
all end responses should call end_response, similarly from inputtags

- should recoginize grade target

- added sig figs to parameters that are passed along
- can now get coorrect responses
- fixed caparesponse to create the proper data bits

Metadata inclusion and syntax fix (missing ";" in structuretags)

- changed Apache::inputtags::part from @ to $, parts are not allowed to nest

- reindent code

New parameter format for metadata

Now also stores away default value for metadata

The responseparam tag now has output for target meta, which will be used
by the RAT and lonparmset to determine which parameters (in addition to the
standard ones) the problem has. The output should be
<parameter_part#_name display="description">type</parameter_part#_name>

- now using name and type instead of type and value

- added <responseparam>, fetches rat set parameters for the question

- lastest for response tag to work

- used to define the response tags