Groups.io: Changelog

Subscribe via RSS


March 9, 2018

  • BUGFIX: Fix redirect to login when trying to view the private guidelines of a group.

March 2, 2018

  • CHANGE: Removed the list of public subgroups from the group home page, as it's a duplicate of what is shown in the Subgroups tab (which also has a better display of them).
  • CHANGE: Added more 'Sent from my ' lines to automatically strip.
  • SYSADMIN: Upgraded the second web server for more capacity.
  • SYSADMIN: Moved one of the two Groups.io DNS servers to a different data center, for redundancy in case our main data center goes off-line.
  • INTERNAL: Moved the final part of the system, messages to karld, off the Redis-based queue and onto a nsq-based queue.
  • BUGFIX: I mispeled the word committed in several places.
  • SYSADMIN: Upgraded the backup database instance to double the memory/disk because we ran out of disk.
  • BUGFIX: There was a deadlock when two users viewed the same group calendar at the same time.
  • BUGFIX: Editing one event in a repeating event series created extra events.
  • NEW: 'All Posts By This Member' in group profile pages.
  • NEW: When a user changes their email address to an address taken by an account with no subs (or subbed just to updates@), we now let them take over that email address.
  • NEW: Moderators can now change members' email addresses.
  • NEW: Moderators can now edit members' signatures.
  • CHANGE: More updates to the Yahoo Group transfer process after I learned that many groups do not accept -subscribe email requests.

February 24, 2018

  • CHANGE: Added meta headers to the maintenance page to hopefully prevent caching.
  • NEW: There's now a status page for the service, at http://status.groups.io
  • CHANGE: Completely revamped how Yahoo groups are transfered, adding a new web-based system to specify the groups to transfer and to track the progress of the transfers.
  • CHANGE: Reduce the max size of a group name to 32 characters to avoid potential DNS issues.
  • BUGFIX: For a specific PDF, pdftotext would hang during processing. This would cause our search ingestion to time out. Now we kill that process if it takes too long.