New enhancements 🛠️

✅ UI standardization

Continued refining titles, links, and buttons to align with overall Neeto UX guidelines.

New features and enhancements 🚀

✅ NeetoCourse reboot begins

We’ve started streamlining the course-building experience, removing legacy complexity built for BigBinary Academy. Most updates this cycle support this new direction.

✅ UI consistency updates

– Improved the empty state when no courses are listed to match other products.
– Removed old Lesson/Exercise pills from the course structure dashboard.
– Updated various UI texts for clarity.
– Renamed “slug” to “URL” across courses, chapters, and pages.

✅ Simplified course creation

Manual course creation is now the default. GitHub integration option has been removed.

✅ ActiveRecord cleanup

Resolved remaining issues flagged by active_record_doctor.

Bug fixes ⚒️

✅ Judge0 timeout handling

Fixed an unhandled timeout error in the Judge0 API for code submissions.

✅ Elasticsearch seeding fix

Implemented retry logic to fix occasional failures in Elasticsearch index creation.

Enhancements:

• We upgraded GitHub integration to have better handling and communication back to the admin when a GitHub app connected to a course is disconnected from GitHub.

• We added Plausible integration on course pages for Bigbinary Academy to track relevant metrics.

Bug fixes:

• We fixed a few more ActiveRecord::InvalidForeignKey errors during incineration introduced into the app when we added all missing foreign keys as part of active_record_doctor audit.

Bug fixes:

• The /progresses client side API endpoint for courses was giving timeouts every now and then. We made a conditional query optimization in result generation when students are logged in and fixed it.

• NeetoQuiz embeds were not working as page component blocks on newly created pages because of an issue with parsing that got introduced into the system when we moved to block based UI. We fixed this issue.

• Since adding a bunch of missing foreign keys to fix active_record_doctor issues from M32, a few issues sprung up where the cascaded destroys started failing when trying to incinerate/destroy some of the resources. We manually added an on cascade: delete for such foreign key relations.

• A few other minor issues came up because of added foreign keys, that required specific exception handling in places where duplicates records creation was not considered earlier. These instances were fixed.

Enhancements:

• Dealt with all brakeman warnings present in the app.

• Fixed all issues reported by active_record_doctor.

Bug fixes:

• Fixed issues with GitHub integration not working correctly in staging.

• Fixed issues that came after upgrading dompurify without having updated its usage. This included fixing a frontend error which was causing internal links on pages to not work on fresh loads.

Bug fixes:

• We found the /services_health_check Endpoint giving timeout errors again despite fixes last time. We debugged and found that for the EUI check, if the EUI is down, it might take 60 seconds before returning a timeout. In general, we decided to fix this for all the apps, and we added an option to add a timeout for such additional services API checks in neeto-commons-backend. Now, any app with an additional service to check can add a timeout; failing that, the API request will fail fast, not triggering NeetoCommonsBackend::Failures.

Upgrades:

• We migrated the packages-upgrade script in neeto-course-site to NeetoCI from github-actions.

• We fixed final set of issues with 404 URLs from academy.bigbinary.com, by setting up the right redirection logic trickling down to chapters and pages, for the courses that don't exist anymore.

• We fixed an issue with evaluation of Promises in JavaScript where despite the user getting the right output, the code was judged not passed. We fixed some edge cases in our evaluation logic to fix this.

• Multiple issues over the last few weeks, related to students not seeing their progress correctly tracked in courses were resolved this week after a little observation. We fixed it by adding an index to avoid possible duplicates, and also by considering the logic that some progress records may be outdated because of the pages/chapters themselves having been unpublished/removed from the content.

• We made the code diffs that get displayed when a student's code doesn't pass the evaluation, a little more informative and highlighting what needs to be fixed.

• We had been observing the /services_health_check endpoint itself timing out and causing honeybadger issues over the last few weeks. On observation, we found that if Elasticsearch is unresponsive, the underlying health check command waits for a long time and causes the entire health check itself to throw a timeout error. We fixed it by adding a timeout to Elasticsearch check at neeto-commons-backend level, so it fails fast if its not responding. This would also work for all other apps that have Elasticsearch in their stack.

• This milestone mostly went into Honeybadger bug fixes. Also, a minor bug in React codeblocks, related to preview failing when an onChange function is added in student code, was fixed.

• We did one final attempt to fix timeout-related Honeybadger issues on the SITE API. However, some of them were re-opened by Honeybadger later. This led us to believe that the issue is not on the app/query level, but something else. On debugging we found that Postgres memory is hogged up almost all the time. We increased the PG memory and have been observing it over the last few days. What we observe, will decide our course of action.

Let's go through what we got done in Milestone M28 (Jan 13 - Jan 26).

• We have had migrations from exercise to challenge and then back to exercise over the last year in the page slugs of exercises for Bigbinary Academy. However, each such change had come with adding redirections to not let existing links break or SEO fail. When we reverted to exercise recently, although the github integration sync took care of adding redirections, older redirections don't make sense anymore. We wrote a script to cleanup the redirections and ensure new ones are in place, and we ran it.

• We found more edge cases in async code evaluation in Javascript and added a fix for one of the edge cases using our existing AST based solution.

• We also had issues with learners tampering with the fixed JS objects (like Array, Number etc) in code and evaluation breaking because of it. A relevant fix that prevents against it was added.

• Earlier, for test cases based JavaScript exercises, we did not support evaluation of async code. This feature was shipped.

• Minor refactoring across the codebase against the ruby best practices audit done earlier, was shipped.

• Long running courses homepage revamp was finalized and shipped this milestone. Course homepage now shows data like number of chapters, hours to complete, projects and prerequisites

• A weird bug where the progress percentage for non-signed in folks sometimes showed over 100%, was fixed. This was because of changing chapters data on the backend but no updates to progress stored in localstorage for non-signed in users. We have a refresh progress button now which does the cleanup.

• Mobile responsiveness for the solution modal in exercises was improved.

• There was a need of removing the run code button on subjective exercises in HTML/CSS/JS challenges on case-by-case basis. We shipped a hideRunCodeButton config on codeblock to support it.

Refactoring/upgrades:

• DB/DB queries analysis done using ruby-pg-extras gem and action items created.

• Refactored some backend code to match Rails best practices standards we follow.

• NextJs on EUI upgraded from 14.2.13 to 14.2.15

• Header component on the EUI was extracted from multiple pages and made standard.