8th of November 2026 (patch release)

This article contains this patch release's bug fixes.

Bug fixes

• Assignment templates are now sorted alphabetically
  Previously, the "Assignment template" list on the new assignment page showed the most recently created template first. The list now shows templates in alphabetical order, ignoring capital letters.
   

• Fixed an issue where some assignment attachments showed "No preview available"
  Some attachments on assignments showed "No preview available" in the digital test instead of the file. This happened to attachments that a teacher had renamed in Ans before 16 August 2026, and to every copy of an assignment with such an attachment.

  On 16 August 2026, we moved all assignment attachments to a new way of storing files. A renamed attachment lost its file type, for example ".pdf", in that move. Without the file type, Ans did not know how to show the file.

  The file type is now added back to these attachments. They show their preview again, in the original assignment and in all its copies. The name of the attachment that teachers and students see stays the same.

  For a small number of attachments, Ans could not find the original file type. If an attachment still shows "No preview available", remove it and upload the file again.
  
   

• Fixed an issue where plain text in LaTeX equations appeared larger than regular Ans text
  Previously, plain text within LaTeX equations did not match the font size of regular Ans text, making the text appear disproportionately large. The text now renders at 90% of the equation's font size, so it no longer looks larger than the Ans text next to it.
   
• Fixed an issue where text with a backslash before a quote in a code block got copies of the rest of the text inserted
  Previously, saving rich text with a code block that held a backslash followed by a quote, such as the '\'' that IntelliJ writes in generated toString() methods, inserted the rest of the text into the code block. Each autosave repeated it, so a student's answer grew with copies of itself and overflowed its box during the exam and in grading. Such text is now saved as typed.
   

• Fixed an issue where unsupported question types in QTI exports were not clearly distinguished from exported questions
  Previously, when you exported questions from a question bank to QTI, the export log showed a question type that QTI does not support in the same grey as the questions that were exported, so it was hard to see which questions were left out.

  These entries now appear in yellow with a warning icon, like the warnings in other question imports and exports. The questions that were exported still appear in grey.
  
   

• Fixed an issue where synchronizing a taken digital test could remove questions
  After moving a timeslot to the future or removing it, a digital test that students were still taking could be synchronized as if nobody had taken it yet. This could, for example, remove questions from a test that was in progress.

  Now, a digital test counts as taken as soon as any student has an attempt in progress or has submitted, whatever its start time. From then on, synchronizing only updates grading-related settings (such as the grade calculation, points and rubrics). It no longer changes the questions themselves.
  
   

• Fixed an issue where students were sent out of a digital test when Schoolyear did not respond in time
  Previously, when a large number of digital tests started at the same time, Schoolyear could respond too slowly to the check that confirms a student is in the Schoolyear browser. If this happened several times within a few seconds, Ans sent every student of that school to the home page for about 30 seconds, with the message "Please restart the Schoolyear session to resume the assignment".

  Students who already entered the digital test through Schoolyear now stay in the digital test while Schoolyear does not respond. Students who did not enter through Schoolyear are still sent away, and so is a student whose session Schoolyear rejects.
  
   

• Fixed an issue where annotated text highlights were not visible in publications with Schoolyear
  Previously, when a publication required Schoolyear, students who opened their answers in the Schoolyear browser saw the annotations in the annotations panel, but the text the reviewer highlighted in their answer to an open question was not marked. The highlighted text is now marked in the answer, the same as outside Schoolyear.
   
• Fixed an issue where the content of an annotation overlapped with the format button
  Text in the comment editor of an annotation now wraps before the format button in the top-right corner of the editor. Long comments no longer run underneath the button, so the button stays visible and clickable while you type.
   

• Fixed an issue where pages after the first could not be annotated or scrolled in a written assignment
  Previously, choosing "Page Scrolling" in the PDF viewer's Tools menu (>>) showed one page at a time. Scrolling did not move to the next page, and annotations could not be placed on most pages. The viewer remembered the choice for every scan made from the same template, so it applied to all results of an assignment and to other assignments.

  PDFs now always open with vertical scrolling, and the Tools menu no longer has a "Page Scrolling" option. Horizontal and wrapped scrolling are still available.
  
   

• Fixed an issue where clicking a link in a criterion comment selected or deselected the criterion
  Previously, clicking on a link in a criterion comment on the grading page selected or deselected the criterion. The link also did not open.

  Clicking now opens the link, and the criterion is no longer (de)selected. The same applies to links in a criterion's description and in comments on "None of the above".
  
   

• Fixed an issue where the round to setting was not always enforced
  In the past, rounding was not enforced in the mark calculation settings. With the release of October 18th, we changed this recommendation to an enforcement. For example, if the rounding is set to 1 decimal, but the pass mark is set to 5.55, the pass mark gets updated to 5.6.

  However, only the mark calculation settings form rounded the minimum mark, maximum mark and pass mark to the selected rounding. Marks set through the API, copied, created from a template, or by synchronising a copy with its original were saved as entered.

  The minimum mark, maximum mark and pass mark are now rounded every time they are saved, however they are set. With "Cut-off Score", marks are rounded down. With "Table", the pass mark is rounded. With "None", nothing is rounded.

  Changing only the rounding now also updates the marks students see. Previously, with "Formula" or "Table", a student with a mark of 5.7 kept seeing 5.7 after the rounding changed from "Round to one decimal" to "Round to whole numbers", instead of 6.

  A pass mark above 0 that would round to 0 is raised to the smallest step of the rounding, for example 1 with "Round to whole numbers". Assignments that already have results keep their current marks until their mark calculation settings are saved again.
  
   

• Fixed an issue where numeric answers were not graded after Error Carry Forward was switched off and on quickly
  Switching "Error Carry Forward" off and on again in quickly in the assignment settings of a digital test could leave numeric answers without a score. Both changes were processed at the same time, and the variables used to grade each student's answers were removed.

  Changes to "Error Carry Forward" are now processed one at a time, using the setting's current value. Answers whose variables were removed are graded against the question's original variables, with errors carried forward from the student's earlier answers.
  
   

• Fixed an issue where the tooltip for the duration histogram took a long time to appear
  Previously, hovering over a bar in "How much time did participants take?" on the insights overview page showed its tooltip only after a delay of a second or more, and the delay started again on every bar.

  The tooltip now appears as soon as the pointer is on a bar. It shows the interval and the number of participants in it. The bars in "How did participants perform?" show their tooltips without the delay too.
  
   

• Fixed an issue where the Excel results export only showed the submission date
  Previously, the "Collection date" and "Submitted at" columns of an Excel (.xlsx) results export showed only the date, while the CSV export showed the date and the time.

  The Excel export now shows these columns as date and time, in the format yyyy-mm-dd hh:mm:ss. The cells stay dates in Excel, so they can still be sorted and filtered.
  
   

• Fixed an issue where questions in the question bank showed 0 for "Times taken"
  Previously, after students took a digital test, the question bank could still show 0 for "Times taken" and an empty "Last taken" for its questions. Their "P-value" could also stay empty. This happened the first time statistics were calculated for a question: several calculations ran at the same moment, and only the first one was saved.

  Every calculation is now saved. Questions that were affected are recalculated, so their "Times taken", "Last taken" and "P-value" columns show the correct values.
  
   

• Fixed an issue where searching for question banks returned no results for matching titles
  Previously, the search field on the "Question banks" tab of an employee's page only searched the 250 most recently updated question banks. Searching for an older question bank showed no results.

  The search field now searches all of the employee's question banks. The search term is added to the page URL, so a search can be shared or reloaded.
  
   

• Fixed an issue where background jobs logs showed "The Application" instead of the user who started the job
  When a user started work that runs in the background, such as synchronising an assignment with its question bank assignment, the log showed "The Application" as the one who deleted attachments, integrations, timeslots and other records in that work.

  The log now names the user who started it. Work that no user started, such as emptying the trash or an LMS sync, still shows "The Application".
  
   

• Fixed an issue where questions with images were missing from assignments created through the API
  Previously, a course assignment created through the API from a question bank assignment could be missing questions. This happened when a question contained an image uploaded by a user who has since been removed. The assignment was created without those questions, and no error was shown.

  These questions are now copied into the assignment. Background tasks started through the API now run as the user the API token belongs to. Changes they make show that user in the logs instead of "The Application".

  Missing questions in assignments created this way before the fix are added to the assignment, unless students have already taken it.