• v2.0.0-rc6 5aabcbf863

    v2.0.0-rc6
    All checks were successful
    release / appimage (push) Successful in 3m26s
    release / rpm (push) Successful in 5m39s
    build / check (push) Successful in 10m40s
    Pre-release

    Ghost released this 2026-10-01 00:50:47 +00:00 | 0 commits to master since this release

    An open window used to spend your whole daily deSEC budget doing nothing. It
    sent over three thousand requests a day while nobody touched it, which was
    enough to be refused by the API and locked out for the rest of the hour. An
    untouched window now sends two.

    Fixed in 2.0.0-rc6

    • Leaving the application open no longer exhausts your API limit. A
      connectivity check ran every sixty seconds and cost two requests each time,
      whether or not anyone was using the window. It is gone: an untouched window
      sends 2 requests a day instead of 3,072, and one in constant use about 96.
    • The window stops syncing after fifteen minutes with no input, and resumes
      the moment you touch it.
      Clicking or typing refreshes everything at once
      rather than waiting for the next scheduled sync.
    • Recovery from a dropped connection takes thirty seconds instead of fifteen
      minutes.
      A failure now retries on its own short schedule, and backs off
      rather than hammering an API that is already refusing.
    • Opening a profile sends two requests instead of four. It was asking the
      same endpoint twice and throwing one answer away.

    Technical Improvements in 2.0.0-rc6

    • The sidebar has an Idle connection state, so a window that has
      deliberately gone quiet does not read as a broken sync.
    • 2244 tests across 88 files, up from 2219 across 87
    Downloads
  • v2.0.0-rc5 e1947f48f2

    v2.0.0-rc5
    All checks were successful
    release / appimage (push) Successful in 4m38s
    build / check (push) Successful in 9m34s
    release / rpm (push) Successful in 6m20s
    Pre-release

    Ghost released this 2026-09-30 13:46:29 +00:00 | 3 commits to master since this release

    The package is a sixth of the size it was, and long-running work no longer
    stops because you looked at another page. Nothing here can lose data, but a
    logout used to abandon an import wherever it had got to, which could leave a
    zone half changed.

    Fixed in 2.0.0-rc5

    • The RPM installs 136 MiB instead of 842, and downloads 37 MiB instead of
      227. It was shipping an entire browser engine and a scientific computing
      stack that this application never loads.
    • An import, export or search keeps running when you switch pages. Leaving
      the page used to cancel it, and coming back now shows it still going.
    • Logging out with work in flight asks first, names what is running, and
      stops it cleanly instead of abandoning it mid-write.
    • Import, export and search report into the Log Console, so a run that
      finished while you were elsewhere leaves a record you can find.
    • The format and mode radio buttons are visible on the dark theme and take
      the accent colour when chosen. They were a grey dot on grey.

    Technical Improvements in 2.0.0-rc5

    • A packaging check runs in the suite and fails if a large dependency nothing
      imports gets back in.
    • 2219 tests across 87 files, up from 2175 across 84
    Downloads
  • v2.0.0-rc4 4ab45dadff

    v2.0.0-rc4
    All checks were successful
    release / appimage (push) Successful in 5m46s
    build / check (push) Successful in 8m49s
    release / rpm (push) Successful in 14m41s
    Pre-release

    Ghost released this 2026-09-29 22:50:19 +00:00 | 8 commits to master since this release

    This release is what came of running the code the test suite had never
    executed. Three of the defects it found could reach a live account: an
    unreadable API token was destroyed by the next save, an import could create a
    zone the file never named, and a destructive confirmation could fire twice.

    Added in 2.0.0-rc4

    • A record of which lines of this application the test suite has never
      executed.
      scripts/line_coverage.py reports every executable line that no
      test run reached, and everything below is what it found.
    • A replace or a delete across the account leaves a restore point. Search
      and Replace was the one interface here with no version history; it now files
      one entry per zone it wrote to, naming the records that really changed, with
      any refusals named beside the count.

    Fixed in 2.0.0-rc4

    • A rate limit of 0, which means no limit, was the one setting the 429 backoff
      could never touch
      , so deSEC could refuse request after request without the
      application ever slowing down. From unlimited it now starts pacing at the
      application default and halves from there, while a limit deliberately set
      below the floor is still left alone.
    • A rate-limit refusal froze the window while the backoff waited on the lock
      that paces requests. It no longer takes that lock.
    • An error response whose detail field was not a string escaped the API
      client
      as an unhandled exception on a queue worker thread, instead of being
      reported.
    • A rate-limit refusal with an empty body put an empty message on screen.
    • params was silently dropped on every request verb but GET, contradicting
      the documented behaviour.
    • An API token that could not be decrypted was destroyed by the next unrelated
      save.
      One launch under a different home directory, or a config.json
      restored without its salt file beside it, was enough. The stored token is kept
      byte for byte now, so a later ordinary launch can still migrate it.
    • A failed token migration said nothing useful. The error named no reason and
      fired on successful migrations as well, so an upgrade that lost the key looked
      exactly like a credential revoked at deSEC; those need different actions and
      are reported differently now.
    • The salt file's permissions are corrected on a file that already exists,
      not only on one this application creates.
    • A clock that had gone backwards left deleted records on screen. A
      timestamp ahead of the current time read as fresh, so a timezone change, a
      daylight saving transition or an NTP correction could pin deleted records in
      the pane and let the delegation check trust a parent copy it had been told not
      to trust.
    • An import could create a zone in the account that the file never named. A
      file that does not name its zone fell back to a real, registrable domain name
      and created it; that is refused now, before any request goes out.
    • A djbdns import could create the wrong zone, taking the zone to be the
      first record's name minus its first label, so a file whose first line is the
      apex of example.com produced the zone com.
    • A merge import that wrote records reported writing none. One record
      missing its TTL abandoned the counts for the whole file; a bad record is named
      and counted now, and the rest of the file still goes in.
    • Exporting a zone stripped the timestamps out of the copy held in memory.
      One export with metadata unticked left every record in that zone with no
      creation or change dates until the next sync.
    • A BIND file this application exported could not always be read back. Six
      defects: multi-value record sets came back as duplicates, types longer than
      six characters were dropped outright, fully qualified owners of the kind dig
      and an AXFR produce were refused, a parenthesised SOA was shredded into bogus
      records, inline comments ended up inside record values, and a legal
      three-field line was dropped in silence.
    • A djbdns export could not be imported at all. The reader understood two of
      the six line types the exporter writes, and colons in an AAAA address and in
      every SPF include: broke a colon-delimited format; all six round-trip now.
    • A line no parser could read is named with its line number rather than
      dropped in silence, in all four formats, and the count reported is what the
      file actually holds.
    • The import documentation was wrong in three places: it required an
      $ORIGIN the BIND reader no longer needs, it listed subname as required for
      JSON and YAML when the apex is deliberately the empty string, and it said
      nothing about unreadable lines being reported.
    • The lines a file could not be read from are shown before you confirm the
      import, not after
      , three named in full and then a count.
    • The preview's "Original Zone (from file)" line could never appear. It shows
      what the file said against the target you chose now, so a deliberate retarget
      can be told from having opened the wrong file.
    • Every destructive confirmation could be answered twice. Deleting a record,
      a zone, a token, a policy, a bulk selection, a whole account's worth of
      matched records or the cache all go through one drawer, and it stayed live for
      the length of its closing animation. One confirmation is one action now.
    • A confirmation asked for without a second step had no gate at all, and
      every click fired.
    • Escape after confirming made the drawer answer twice, cancelled as well as
      confirmed, which put the Import page back to idle over a run it had already
      started.
    • Deleting the same token twice while the list refreshed reported a failure
      for a token that had already gone. Token deletes carry the per-item guard that
      policy deletes were given, keyed per item, so two partly overlapping
      selections still delete every token in them.
    • Search and Replace's destructive buttons had no re-entrancy guard, so a
      second click enqueued the same records again, turned rows red for deletes that
      had in fact succeeded, and overwrote the first run's summary.
    • Clearing the Queue page's history could swallow a callback that had not
      fired yet
      , which left whatever was waiting on it waiting for ever. The
      dispatch no longer looks an item up in a table that the clear prunes.
    • Retry on the Queue page left the item in the state that had failed. It no
      longer starts with its retries spent, no longer shows the previous failure's
      response in the detail panel, and no longer comes back after a restart.
    • The Pause button on the Queue page followed only its own click, so it
      still offered to pause a queue the window had already paused for offline mode
      or a rate-limit cooldown.
    • A typo in the TTL filter widened a Search and Replace to the whole account.
      A TTL that is not a number was ignored in silence, on a page whose next two
      buttons are Apply Changes and Delete Selected; it is reported now.
    • A rename that could not delete the old record was reported as a clean
      rename.
      It counts as a failure now and names the record left behind.
    • Deleting a zone left a stale refresh running behind it that came back 404
      for the zone just removed.
      Pending work for a zone is cancelled before the
      delete is sent as well as after it, with other deletes left alone.
    • Confirming the same zone delete twice reported a failure for a zone that had
      just been deleted.
      A delete cancelled by an earlier delete is no longer
      rendered as an error, while a cancel asked for from the Queue page still is.
    • A zone delete on a window with no queue said the zone was deleted and left
      it in the list
      , with Delete Zone still enabled and pointing at it.
    • A zone the server named differently from the way it was typed never got a
      padlock.
      The name in the create response is what gets registered for the
      delegation and DNSSEC check now, so the verdict lands on a row that exists.
    • Refusing to add a zone because the application is offline was completely
      silent.
      A panel opened while the connection was up and submitted after it
      dropped slid away with the typed name and left nothing on screen; it says why
      now.
    • Double-clicking a zone while offline raised "Editing zones is disabled in
      offline mode"
      for a gesture that only reads.
    • The zone counter dropped its units whenever a filter was typed, reading
      "Showing 1 of 2/100" instead of naming zones.
    • The countdown under a freshly created token was a label, not a clock. A new
      token went to the clipboard for thirty seconds under a string that only
      claimed to count down. It is two minutes now, ticking once a second in one
      line under the token, in amber and then red for the last ten seconds.
    • Closing the token dialog clears the clipboard early, and the dialog says so
      while the token is still on screen.
    • The clock never stops while the token is still on the clipboard. A
      clipboard it cannot read keeps it running, and it ends only on evidence that
      the token has gone: other text, or another application holding content it will
      not hand over as text.
    • Copying a token twice started two clocks and the shorter one won, a timer
      could fire into a destroyed dialog, and an empty token offered a Copy that
      would have cleared an unrelated clipboard.
    • Deleting a token's last RRset policy makes that token unrestricted, and
      nothing said so.
      It is said in the confirmation, while there is still
      something to change, and again from the refreshed list afterwards.
    • A refused policy delete names which policy it was, where three selected and
      one refused gave three identical lines.
    • The same policy could be deleted twice while its row waited for the
      refresh
      , and the second 404 was reported as a failure over a restriction
      that had already gone.
    • An emptied policy list was not cached, so the next cache-first paint showed
      a restriction that no longer existed.
    • A zone with nothing known about it was treated as settled and left for a
      week.
      A zone whose first check failed comes back for a recheck in hours now,
      which is what the unsettled interval is for.
    • Maximising the Add Zone wizard stretched its four mode tiles down the window
      and crammed them into its left half.
      The grid takes one, two or four columns
      according to the room it has, the slack sits at the bottom of each card, and a
      line of text stays about the same length however wide the window is.
    • Ctrl+F could only ever reach the zone search field, never the record one.
      Both are in the cycle now.

    Technical Improvements in 2.0.0-rc4

    • Three more invariants are enforced by detectors that run in the test
      suite
      , numbered I10 to I12: a module that owns state outliving a window may
      not ship a member a user can reach that no test has ever executed, one
      structure may not be written out as two literals, and a test stand-in may not
      be incapable of taking a call the source really makes.
    • The DNSSEC and delegation status entry is built from one defaults table,
      where it had been written out by hand in seven places.
    • A test file that passed and then died is no longer reported as green.
      scripts/test_token_page.py printed a clean summary and then exited on a
      signal; the runner reads the exit status now and reports a crash as a crash,
      and the crash itself is gone, each test closing and deleting its top-level
      widgets before the interpreter shuts down.
    • A wizard run asked for a read and a snapshot of every zone it aimed at,
      including the ones the API refused.
      It uses the set of zones actually
      written, which it had already worked out. A stand-in for the API client in
      that area could no longer take the call the source makes, and is fixed with
      it.
    • A cache test forces its failure by injection rather than with a read-only
      directory
      , which root is not subject to, so it no longer passes locally and
      fails in the CI container.
    • 2175 tests across 84 files, up from 1634 across 64
    Downloads
  • v2.0.0-rc3 a090693ca5

    v2.0.0-rc3
    All checks were successful
    release / appimage (push) Successful in 9m24s
    build / check (push) Successful in 10m43s
    release / rpm (push) Successful in 31m46s
    Pre-release

    Ghost released this 2026-09-26 12:58:50 +00:00 | 32 commits to master since this release

    A qfluentwidgets table paints its highlight from row numbers of its own, so a
    sort or a background refresh could leave the highlight on one row and the
    selection on another, and a delete then destroyed records, zones or API tokens
    other than the ones highlighted. Treat a bulk delete on rc1 or rc2 as unsafe.

    Added in 2.0.0-rc3

    • A route to clearing the cache, in the sidebar directly above Sync. The
      action, its two-step confirmation and the re-sync behind it have been there
      since the menu bar, but nothing in the Fluent interface could reach it, and
      giving it a route found it broken: a clear left the records cache in a state
      where the first zone opened afterwards failed, and nothing was cached again
      until the application was restarted.

    Fixed in 2.0.0-rc3

    • A bulk delete destroyed records the highlight said were not selected.
      Sorting a column, or a background refresh landing, moved the rows while the
      painted highlight stayed on the old row numbers, and the delete followed the
      paint. A record is held by its name and type within one zone now, so nothing
      that slides into a deleted record's row number can be selected in error.
    • A token delete after sorting the table destroyed tokens other than the ones
      highlighted
      , and an API token cannot be recovered once deleted. Tokens are
      held by id now and policies by policy id, and the invisible second load of
      that page no longer takes the selection with it.
    • Select All on the Export page selected every zone, highlighted none, and
      Export exported nothing at all
      , while typing in the filter box emptied the
      selection outright. The same defect left a stale highlight on the DNS page
      after a sync and on the DNSSEC page; all three hold their selection by zone
      name now, and a zone that has left the account is not put back.
    • Search and Replace acted on rows nobody had chosen. Its table is rebuilt
      from empty on every search, so a second search left the highlight standing
      over whatever was underneath; it is held by zone, subname and type now, which
      is the unit Replace and Delete act on.
    • Retry Failed on the Queue page could retry an item that had just arrived.
      Neither of its tables empties itself, so both halves of the selection kept
      their row numbers and agreed with each other while naming items the user never
      picked. They are held by queue item id now.
    • Leaving the History page and coming back took the selection while the
      highlight stayed and Delete Selected greyed out underneath it. It is held by
      zone name.
    • The wizards painted rows that had gone. Filtering the Add Zone wizard's
      domain list left twelve rows highlighted over an empty selection, Remove
      Selected in the Custom Record Set builder drew two rows as picked that were
      not, and a row picked before Re-check in the Create Subdomain wizard stayed
      painted for the rest of the run.
    • Import and export went around the API queue and its rate limiter entirely,
      so nothing appeared on the Queue page and a rate-limit refusal was never
      handled: an import refused early carried on spending requests and then
      reported itself a success. Every request goes through the queue now, with a
      readable action and one of the Queue page's categories, and a run that stops
      counts what it wrote, what was refused and what was never attempted.
    • A finished import left no restore point, and going back to the zone served
      the pre-import records again.
      The completion names the zones the import
      wrote to now, so each one gets a version history entry and is read back from
      deSEC; a run stopped partway still names what it changed.
    • A cancelled import claimed work it had not done, recording the phase it was
      about to start as the last thing written. It names the last change actually
      made now, or says plainly that nothing was written.
    • The Import page read as idle while it was still writing to deSEC. The
      confirmation is a child overlay rather than a modal dialog, so Preview Import
      stayed live underneath it, and a preview started there marked the page idle
      while the import was still writing. The page is locked from the moment the
      confirmation opens until it is answered.
    • A replace wrote whatever the form held at the moment of confirming, not the
      values that were confirmed.
      Find, Replace, Subname and TTL all stayed
      editable behind the confirmation, which never showed the values either. All
      four are captured when the button is pressed now, and the drawer lists the
      changes it is confirming, one line each.
    • Two searches could run at once and the older one painted the table. The
      Return key in a filter field started a second search over the first, so the
      table could show results from a filter the page was no longer displaying, with
      Apply Changes live over them. There is one live search at a time now, and one
      way it stops.
    • The search worker fetched the zone list itself, around the queue and its
      pacing
      , including while the queue was paused for a rate-limit cooldown. It
      reads only the cache now; with nothing cached and the queue paused the search
      is refused and says why.
    • Pressing Search on a profile that had never synced raised an error instead of
      searching.
      Every read of the zone cache on that page goes through one
      accessor now.
    • A confirmation drawer's Delete button rendered as "De". A drawer in a pane
      narrower than its own minimum width was clipped at both ends; the pane is a
      hard ceiling now. Four kinds of Settings card had the same shape, running
      their control past the card's right edge and squeezing the title to make room.
    • A wizard run reloaded the records of a zone it had not touched. Both
      completion signals name their zones now, so the records pane refreshes only
      when the zone being read is one of them.
    • Opening a profile froze the window for a round trip before it had drawn
      anything.
      The API client ran a connectivity check from its own constructor,
      and saving a token ran another. The window makes one check per profile open
      through the queue now, and it respects offline mode.
    • Every sync sent the user back to the first zone in the sidebar, and spent a
      record fetch nobody had asked for. Both zone-loading paths put the selection
      and the cursor back where they were, unless the selected zone has left the
      account.
    • Scrolling the zone list stuttered while a delegation and DNSSEC sweep was
      running.
      The spinner beside the zone being checked announced every tick as a
      change of data, which a list view answers by laying out and repainting the
      whole viewport. It repaints its own rows now, and skips any row off screen.
    • Flipping Debug Mode did nothing at all until Save Settings was pressed, so
      debug mode was never really being enabled, next to a switch in the same group
      that applies the instant it moves. It is live now, raising or lowering the log
      level there and then, and a Save is no part of it.
    • Changing a setting left nothing in the Log Console. The API URL, the sync
      interval, the rate limit, queue persistence, history retention and the theme
      each say one line when they actually change, and a Save that changed nothing
      says nothing.
    • The documentation called each profile's version history shared. It has been
      per profile since 2.0.0-rc2 and the old wording survived in the passphrase
      panel, the roadmap, the README and three reference documents; it says per
      profile now, and names the read-only installation-wide archive beside it.
    • The README is an introduction again. Every section that left landed in the
      document that owns the subject, install, signing, privacy, rate limiting,
      DNSSEC and record management, and What you get now covers the demo account and
      the wizard's reverse DNS mode, neither of which it had mentioned.
    • SECURITY.md states the one fact nothing in the repository stated: every
      valid deSEC token can read every zone in its account, and RRset policies
      restrict writes only.

    Technical Improvements in 2.0.0-rc3

    • Eight invariants are enforced by detectors that run in the test suite,
      discovered by the test runner and checked on every push, with
      doc/STYLE-GUIDE.md carrying all eight. Five suppressions were found still
      switched on for defects already fixed, and an entry can now record the guard
      text it rests on, so a fix that deletes the guard cannot be waved through by
      the review the guard was the reason for.
    • The shared code survey tells two same-named signals apart. Twelve signal
      names in src/ are declared on more than one class, and a question about one
      of them used to be answered with every site that connects to any of them; a
      site the resolution cannot reach now says so. A companion script runs the CI
      check job locally in the container image the runner uses, against a clean
      export.
    • Twenty-one unreachable methods removed, among them ten leftovers from the
      menu bar the Fluent sidebar replaced and five configuration setters that wrote
      to the in-memory configuration without ever saving it. A test fails if any
      removed name reappears.
    • A test harness was leaving a git repository behind for every test it ran. It
      takes its own directory away now, after waiting for the worker that runs git
      to stop.
    • 1634 tests across 64 files, up from 1394 across 46
    Downloads
  • v2.0.0-rc2 ab78e1a299

    v2.0.0-rc2
    All checks were successful
    release / appimage (push) Successful in 5m37s
    build / check (push) Successful in 7m20s
    release / rpm (push) Successful in 15m7s
    Pre-release

    Ghost released this 2026-09-22 17:38:43 +00:00 | 87 commits to master since this release

    Version history becomes per profile: two profiles holding the same domain
    shared one installation-wide history, so a restore in one account could write
    the other's records over it. A record edit or a bulk delete confirmed after
    switching zones also acted on the zone that was selected by then.

    Added in 2.0.0-rc2

    • Every profile now has its own version history. Two profiles holding the
      same domain shared a single history, so restoring the oldest version of
      example.com in one account could write another account's records over it.
      The old repository is opened read only and listed under Earlier shared
      history
      with Restore disabled and the reason beside it, rather than being
      merged or deleted.
    • An account summary on the Settings page: the profile, the account email,
      how many domains it holds, how many of those are verified, and how many record
      sets are cached. The verified count is red when a domain is unverified,
      because deSEC stops raising the domain limit while one is, and a token without
      the manage tokens permission is now told that rather than being told the
      account has not synced yet.
    • Qt's own messages now reach the Log Console. The Wayland and X11 platform
      plugins, QSslSocket and QProcess log through a channel separate from Python's
      logging and previously reached only the terminal, which is nowhere at all for
      anyone starting this from a desktop menu.
    • A demo profile says what it is in the Log Console, as the first line there:
      the data is generated, and the live delegation and DNSSEC checks do not run,
      with every padlock derived from the stored records. "Retrieved N zones from
      API" no longer appears on a demo profile either.

    Fixed in 2.0.0-rc2

    • Help tooltips turned up anywhere on the screen under Wayland, nowhere near
      the question mark they belong to, with an opacity warning repeating on the
      console. A tooltip is an ordinary child of the window now, clamped to that
      window's own edges, which behaves identically on Wayland, on X11 and
      offscreen.
    • A one-word help tooltip was drawn as a wide slab sitting to the left of its
      icon.
      The label is sized from its own text now, capped where it always was
      so a long tooltip still wraps exactly as before.
    • A help tooltip's background was black on some pages and right on others,
      depending on whether an ancestor carried its own stylesheet. Every help icon
      uses the themed tooltip qfluentwidgets ships.
    • An apostrophe in a help tooltip was painted as '. Ampersands and
      angle brackets are still escaped.
    • Every version history entry described a change it did not contain. The
      snapshot was taken from the record list the widget was already holding, which
      is the state from before the write, and where that list matched what was
      already committed the entry was skipped in silence. The commit is taken from
      what the API returns on the refresh that follows every mutation now, keyed to
      the zone it names.
    • A zone emptied of all its records could not be recorded at all, which is
      the state a user is most likely to want back.
    • A bulk delete left no version history entry whatsoever. It files one entry
      for the batch now, naming up to three records and past that a count with a
      sample, and a batch the API refused in part records the accepted count with
      the refusals named beside it.
    • A record opened for editing while its zone was still loading was saved into
      the wrong zone.
      Rows are dropped the moment the selection changes, an answer
      for a zone already left is discarded, and every row carries the zone it was
      built for.
    • A bulk delete confirmed after switching zones deleted from the wrong zone.
      The confirmation is an overlay rather than a modal dialog, so the sidebar
      stays live while it is open; the zone is captured with the rows now.
    • Everything a queued mutation did afterwards followed the user rather than the
      write.
      Clicking away while the queue waited left the changed zone serving its
      old records on the next visit, spent a request re-reading a zone nothing had
      happened to, and left the history entry to be absorbed by the next change made
      there. Every mutation names the zone it wrote to, and that is the zone
      invalidated and read back.
    • Two bulk deletes running at once shared one set of counters, so the zone
      that was emptied got no history entry, the zone that was not touched got
      several, and the counts did not match the batch. Each batch owns its own state
      now, and a second is refused while one is running, said out loud in a
      notification and in the Log Console.
    • A multi-row Delete went past the API queue and its rate limiter entirely
      when it was started with the Delete key, and the toolbar button's guard
      re-enabled itself on the next change of selection. Both ways in run one
      implementation behind one guard now.
    • Saving a token could write its name and permissions onto another token's row,
      after which deleting that row deleted the wrong token.
      An API token cannot be
      recovered once deleted. The answer is matched to the row carrying the id it
      was saved for and does nothing when no row carries it, and a token's policy
      panel now carries the token it was opened for.
    • A refused policy list was displayed as a token having no policies, which in
      deSEC means unrestricted, the opposite of the truth for anyone deciding
      whether that token is safe to use. A refusal is reported as a refusal now.
    • A token whose default policy could not be created was still reported as
      created
      , which leaves an unrestricted token with write access to every zone.
      That failure is reported now, stays on screen until it is dismissed, and says
      what state the token is in and what to do about it.
    • Leaving Search and Replace mid-search froze the window for two seconds and
      then carried on searching
      , with its callbacks still live, so one landing
      afterwards started a fresh search on a page nobody was looking at. Import and
      export instead terminated their thread, which for an import can be between two
      writes; all three cancel between items now and report what was actually
      written, and a cancelled merge import is no longer reported as nothing
      written.
    • Text on the Queue page was unreadable on the light theme. It used one set
      of colours for both themes where every other module pairs them, and it reuses
      the record editor's pairs now, each checked against the contrast floor that
      applies where it is drawn. Links stay blue rather than taking the new red
      accent, and one pair serves the About page and both drawers.
    • Toasts in the profile window cut their own message in half. Two wrapping
      mechanisms were fighting and nothing recomputed the height, so only a one-line
      message was ever safe. A bar sizes itself to its text first now, caps its
      width only when that does not fit, and breaks a long unbroken profile name
      across lines instead of pushing out past both edges of the window.
    • The raw body of a failed API response was printed in the log, where a JSON
      error object reads as line noise to someone looking for what went wrong. The
      line names the subsystem, what was attempted and the cause, a 404 names the
      URL that was not found, and the body moved to a debug line.
    • A zone created or delegated by a wizard waited up to ten minutes to be
      checked.
      All three wizard modes ask for the check that the plain Add Zone
      path has always asked for now, and a new zone is registered rather than merely
      enqueued, so its verdict is not discarded for being about a zone the checker
      has not heard of.
    • Nine pixels of misalignment on five pages, wherever a help button faced a
      bare caption: Tokens, Queue, History, Search and Replace, and Import and
      Export

    Technical Improvements in 2.0.0-rc2

    • Every record mutation read its zone back twice, one wasted request on every
      save, edit and delete, against a daily rate limit this application does reach.
      The signal carries its zone now, and carrying it is the promise that the
      emitter has already invalidated that zone and read it back.
    • A restore read the account on the GUI thread, and around the rate limiter
      every other call goes through. It is queued now, with the write built and
      enqueued from the read's callback, so the read still precedes the write.
    • The record selection is cleared deliberately when the zone changes, rather
      than as a side effect of the table being rebuilt, which is what Delete
      Selected had been resting on. Removed alongside it: two guards reading
      attributes that are assigned nowhere in this application, and the two methods
      under them.
    • The guard that keeps raw response bodies out of the log could not see most of
      them
      , including every queue callback in this application, which takes its
      payload as a parameter. It catches all nine variants now.
    • Geometry assertions no longer assert which fonts are installed. The ten
      text-derived comparisons allow two pixels, which is useless to a real defect
      and sufficient for a font; pixmap arithmetic is left exact.
    • 1394 tests across 46 files, up from 976 across 35
    Downloads
  • v1.8.0-beta d46a5b5cb9

    v1.8.0-beta
    Some checks failed
    build / check (push) Has been cancelled
    release / appimage (push) Successful in 5m39s
    release / rpm (push) Failing after 7m48s
    Pre-release

    Ghost released this 2026-09-15 18:33:51 +00:00 | 146 commits to master since this release

    A profile window now opens before the main one, so a profile's API token can be
    locked behind a passphrase and only one window at a time can hold a profile.
    The reverse DNS wizard mode arrives here, and with it the fix for a run that
    could write generated PTR records into a zone belonging to somebody else.

    Added in 1.8.0-beta

    • A profile window, shown before the main window every time. The profile
      decides which token, cache and settings the rest of the session is built from,
      and a protected profile has to be unlocked before any of that exists, so
      everything to do with profiles happens there and the sidebar's Profile page is
      gone. Closing the window quits, and there is no cancel to nowhere
    • A profile's API token can be locked behind a passphrase. Scrypt derives the
      key, Fernet encrypts the token and a verifier blob tells a wrong passphrase
      from a corrupt file; set a passphrase when creating the profile and the token
      is never written unprotected even once
    • The window says exactly what a passphrase protects, in the window and not
      only in the documentation: the API token, and nothing else. Cached zone and
      record names, the queue history, the shared version history and the shared log
      stay readable on disk
    • Forgotten your passphrase? There is no recovery, by design, so the escape
      asks twice and then clears the stored token, keeping the profile, its cached
      zones and records and its settings; the old token keeps working until it is
      revoked at deSEC
    • One instance per profile. Opening a profile takes an exclusive lock on a
      file in its directory for the life of the process, so a second window on the
      same profile is refused by name, while two windows on two different profiles
      work
    • Reverse DNS wizard mode. Turns IP prefixes into the reverse zones that
      cover them, 192.0.2.0/24 into 2.0.192.in-addr.arpa and 2001:db8:1:2::/64
      into its ip6.arpa name, and optionally fills each one with PTR records from
      a naming pattern. A prefix that does not land on a label boundary is covered
      by several zones at the boundary below it, and a /25 to /31 is refused by
      name, since sharing a /24 needs RFC 2317 delegation from whoever holds it
    • The wizard's preset library reworked against current vendor documentation.
      70 presets in 18 categories becomes 61 in 16, every preset holding a
      hardcoded third-party address having been checked against that vendor's own
      documentation
    • Six presets added: Stalwart Mail with the MTA-STS trio and DANE, a null MX
      "Disable Email" set for a domain that should neither send nor receive mail, a
      current Google Site Verification, a Bluesky handle, an HTTPS service binding,
      and Apple iCloud+ custom email
    • Every multi-select list now says so. Twelve views accept Ctrl+click and
      Shift+click and not one of them advertised it, so the bulk actions beside them
      read as though they applied to the single highlighted row; a question mark
      beside each list carries the gestures in its tooltip
    • The Variables step groups each variable under the preset that asks for it,
      in the order the presets were picked, instead of one alphabetical run of
      unfamiliar names with no way back to the previous screen
    • A logo, in the four places the application introduces itself: the title
      bar's upper left corner and the task bar entry beside it, next to "Choose a
      profile" in the window that opens first, and above the name and version on the
      About page and in the About drawer. It follows a theme change as it happens,
      and the artwork is chosen by measured luminance rather than by file name
    • One name everywhere: "deSEC Qt DNS Manager". The window title and the About
      page said "deSEC DNS Manager" while the package, the desktop entry,
      setApplicationName and the documentation said the longer form; both windows
      take the name from about.APP_NAME now, so there is no literal left to drift
    • The About page leads with the licence, ahead of the author and the credits,
      where it had been the last line of a panel that has to be scrolled

    Fixed in 1.8.0-beta

    • A reverse DNS run could overwrite somebody else's PTR records. A zone
      planned as "create" that turned out to exist was probed, any answer was taken
      as proof its own earlier create had gone through, and 254 generated PTRs went
      into it. The zone's contents decide now: any PTR record in it means the zone
      is left alone
    • Unlocking a profile with the Enter key did nothing except silently clear
      what had been typed. An open panel owns the Return key now
    • Cancelling a queued item from the Queue page could wedge a wizard
      permanently
      , with Back, Next and Start Over all refused and nothing to do but
      restart. Every enqueued item's callback now fires exactly once, whatever
      becomes of the item
    • Records in reverse zones sorted 1, 10, 11, 2. Names and content sort the
      way a person reads numbers now
    • A profile's cache directory was created shared and then repointed. No data
      crossed between profiles, but the directory should never have existed
    • Error toasts in the profile window ran past the edge of the window and lost
      the end of the message. They are capped, centred and wrap to a second line
    • Confirmation drawers spanned the full width of the window, putting a single
      sentence across roughly two thousand pixels, which read as far heavier than
      the thing being confirmed. They are capped, centred cards now, and logging out
      asks once rather than twice
    • The application had no icon at all, in any window or any task bar. Startup
      asked for a relative path to a file that has never existed in the repository,
      which Qt answers with an empty icon and no complaint

    Technical Improvements in 1.8.0-beta

    • The blocking work is off the UI thread. The DNSSEC status store batches its
      writes instead of rewriting its whole file once per finished check, zone
      snapshots run their git subprocesses on a worker rather than on the thread
      about to repaint, and the paint path no longer copies a dictionary per visible
      row per repaint
    • The records held in memory are bounded at fifty domains, least recently
      used evicted first, where every domain visited used to stay for the life of
      the session. Nothing is lost by an eviction: the file cache still holds it
    • The queue keeps a thousand history entries by default, up from 500
    • Counted nouns go through one helper, so nothing in the interface says "1
      records"
    • No em dash appears in any tracked file, enforced by a test that walks them
      all
    • Every example in the repository is documentation or private address space
      (RFC 1918, RFC 3849, RFC 2606). Real prefixes, reverse zones and domain names
      from a live account had reached test fixtures, docstrings, UI copy and the
      documentation
    • The logo is drawn at the display's device pixel ratio, so a HiDPI screen
      gets the artwork at its real resolution, and the ratio is part of the cache
      key, or a window moved to a display with a different ratio would be served the
      stale pixmap
    • The logo assets ship inside the application, under src/assets rather than
      img/, which is what both the RPM and the AppImage actually package; the
      frozen build reads them through sys._MEIPASS
    • The old logo/ directory is gone and the packaged desktop icon is
      generated from the new artwork, square at 256 px because it installs into
      hicolor/256x256/apps/. A test walks every file path the release workflow, the
      RPM spec and the README point at and fails if one stops existing
    • 657 tests across 24 scripts, up from 524, including new suites for the
      version manager, cache isolation, the memory bound, the selection hints and
      the logo
    Downloads
  • v1.7.0-beta a6bbf8d34f

    v1.7.0-beta
    All checks were successful
    build / check (push) Successful in 2m25s
    release / appimage (push) Successful in 4m11s
    release / rpm (push) Successful in 11m0s
    Pre-release

    Ghost released this 2026-09-13 16:36:57 +00:00 | 208 commits to master since this release

    A chain link now sits beside each zone's padlock, saying whether the parent
    actually delegates the zone to deSEC, and the checks run in the background on
    whatever page you are on. Before this, nothing was checked unless the DNSSEC
    page was open, so a padlock was only as fresh as the last visit there.

    Added in 1.7.0-beta

    • A delegation indicator beside each zone's padlock, a chain link saying
      whether the parent actually delegates the zone to deSEC's nameservers. The two
      findings are independent, so with the padlock alone a zone that had left the
      account read as an unremarkable grey "unsigned"; there are four states, at
      deSEC (joined, green), partly at deSEC (joined, amber), elsewhere (broken,
      amber) and not delegated (broken, red)
    • The delegation costs no extra query in the default lookup method: the
      referral the walk already follows carries the parent's NS RRset. The resolver
      and DNS-over-HTTPS methods send one NS query per zone and read the zone's own
      RRset instead, which catches a domain that has moved away but not a parent
      listing deSEC alongside a stranger, and the tooltip says so
    • Checks now run in the background, a pass every 10 minutes whatever page is
      in front, taking up to 25 zones whose answers have expired, oldest first, with
      the first pass shortly after startup rather than during it
    • A zone is checked the moment it is created, instead of carrying no icon
      until something happened to reach it
    • The zone you select is checked first when its answer has expired, on both
      pages, and costs nothing when it has not
    • Recheck Interval setting: how long a healthy answer is kept, 1, 3, 7
      (default), 14 or 30 days, from a dropdown rather than a number to type. The 6
      hours a broken, undelegated or erroring zone waits is deliberately not
      offered, and changing the interval takes effect at once
    • An icon legend, from the info button beside the domain list: every icon
      next to what it means, in one popup
    • Tooltips lead with the verdict, joining both findings in colour and bold,
      then the evidence, then how old it is in grey. The headline takes the colour of
      the worse half, so a green delegation cannot dress up a broken chain beside it
    • doc/DNSSEC-CHECKS.md, a document of its own for the
      feature: the icons, every trigger that starts a check, how long answers are
      kept, the lookup methods, and the one case the check cannot settle

    Fixed in 1.7.0-beta

    • A tooltip about a zone appeared over the records pane, Qt showing an item
      view's tooltip wherever the pointer is. Tooltips are anchored under their row
      now, in both lists
    • Tooltips used the system's pale yellow styling on a dark window. They now
      follow the app's theme, and the status icons repaint when the theme changes
      rather than keeping the colours they were painted with
    • A zone whose delegation had just been fixed could keep a red padlock until
      the next background pass, while the Chain of Trust card beside it already read
      "NS: in place". Lifting that verdict now asks for the recheck it just made due
    • A delegation finding could outlive what it rested on, in three places:
      lifting an override, the wizard writing a delegation, and a co-hosted zone with
      a leftover DS reading as delegated to deSEC while resolving nowhere. All three
      are settled against the parent's own records now
    • A walk that failed after seeing a referral discarded the delegation it had
      already established
      , which is exactly the lame-delegation case where it
      matters
    • Settings showed three dead dropdowns while the feature was off. The lookup
      method, provider and recheck interval are details of that feature and now go
      away with it
    • A new zone's result could be discarded seconds after it was checked, if the
      DNSSEC page was opened before the zone list finished syncing: the page rebuilt
      its list from a cache that did not know about the zone yet
    • Walking the zone list with the arrow keys could queue the whole account.
      The check now waits for the selection to settle, so only the row stopped on is
      checked
    • The "DNSSEC checks unavailable" warning could appear over an unrelated page
      twenty seconds after launch, with nothing on screen to connect it to. It now
      interrupts only when its own page is in front; the log line and the note under
      the domain list record it either way
    • The broken link's outer round caps were clipped square by the edge of the
      canvas

    Technical Improvements in 1.7.0-beta

    • Each status icon carries a 2x pixmap, so a HiDPI row is painted for its screen
      rather than stretched
    • Entries written before the delegation finding existed are due at once, so they
      fill in after an upgrade instead of waiting up to a week. Absent, not False: a
      check that ran and could not settle the delegation waits its ordinary turn
    • After a batch finds DNS unreachable, the background pass waits an hour and
      tries again on its own. The page-show path still waits to be asked, which is
      right where somebody is present to press Verify all
    • delegation.classify_ns() decides what a set of nameservers means, beside
      DESEC_NAMESERVERS, so this feature and the subdomain wizard cannot disagree;
      dnssec_probe reports the names it found and nothing more
    • The cache stores those names rather than a verdict about them, so a file
      written today still reads correctly if the provider's nameserver list ever
      changes
    • 355 tests across 9 scripts, up from 315
    Downloads
  • v1.6.0-beta 44e565c656

    v1.6.0-beta
    All checks were successful
    build / check (push) Successful in 3m8s
    release / appimage (push) Successful in 5m10s
    release / rpm (push) Successful in 14m59s
    Pre-release

    Ghost released this 2026-09-13 03:16:16 +00:00 | 219 commits to master since this release

    Per-zone DNSSEC chain checks arrive here, off by default. They answer the one
    question deSEC cannot: whether the parent publishes a DS whose digest matches a
    key the zone really serves, anchored to the root. A stale DS after a rollover
    leaves a zone perfectly signed at deSEC and unresolvable everywhere else.

    Added in 1.6.0-beta

    • Per-zone DNSSEC chain checks, off by default. Turn on DNSSEC Chain
      Checks
      in Settings, Advanced, and a padlock appears beside every zone, on the
      DNSSEC page and on the DNS page's zone list. Hovering it explains the verdict
      and says how old it is
    • Three lookup methods, chosen in Settings and never switched
      automatically: the direct walk of root, TLD and authoritative servers (the
      default, which needs outbound port 53), this machine's own resolver, or
      DNS-over-HTTPS through Cloudflare, Google or OpenDNS. The setting takes effect
      immediately
    • Verify all, or Verify (N) for just the domains picked out of the list, the
      same way the DNS page's Delete button already works
    • A DNS-access preflight before each batch, and a breaker. A network that
      answers nothing is reported once, instead of timing out zone by zone; a run
      that collapses after it starts stops itself after eight consecutive transport
      failures and says so
    • Results cached per profile in cache/dnssec_status.json, with a recheck
      policy: broken, undelegated and error zones after 6 hours, secure and unsigned
      after 7 days, dropping to 6 hours when the last check failed. Creating or
      completing a subdomain delegation in the wizard marks that zone due
      immediately
    • Every check narrated in the log console, so a slow or failing lookup can be
      seen rather than guessed at
    • A parent in the same deSEC account is settled by the API, because when both
      zones are served by the same nameservers a leftover DS with no NS records looks
      exactly like a working delegation. The API's answer wins and is cached with the
      result

    Fixed in 1.6.0-beta

    • An empty non-terminal counted as a break in the chain. Labels like
      2.ip6.arpa are not zones and cannot carry a DS, so reverse zones read as
      broken when they were fine. An NS query now tells a real delegation point from
      an empty label
    • A zone could show a padlock it had not earned. A verdict is now stored only
      once the delegation is confirmed, the parent's records are trusted only while
      they are less than 5 minutes old, and a failed read never counts as
      confirmation
    • Switching the lookup method could freeze the window for about 3 seconds and
      still return with the worker thread running; a check in flight is now cancelled
      where it waits
    • A stop requested before the worker thread started was swallowed, leaving a
      worker that would not stop
    • The breaker's error streak leaked between batches, so an earlier outage
      could stop a later healthy run
    • Loading the stored setting looked like the user toggling it, which started
      a sweep nobody asked for
    • Settings overflowed and clipped its dropdowns. The page is rearranged, the
      lookup dropdowns are sized to their contents, the DNS-over-HTTPS provider card
      appears only when that method is selected, and the lookup method greys out
      while checks are off
    • The duplicate Token Manager card is gone from Settings; Tokens has had its
      own sidebar page since the Fluent rework

    Technical Improvements in 1.6.0-beta

    • Queue history no longer rewrites the whole file on every API call. Each
      entry is serialised once, when it enters the history; writes are coalesced to
      at most one every 3 seconds, with forced flushes on stop and on clear; and a
      single response is capped at 32 KB on disk
    • dnspython added to the hash-locked dependencies
    • Four new modules: dnssec_probe.py (the walk, with no Qt and no app state),
      dnssec_status.py (the per-profile cache), dnssec_worker.py (the background
      thread) and dnssec_icons.py (padlock painting and tooltip wording, shared by
      both zone lists)
    • The feature is documented in doc/ARCHITECTURE.md, doc/CONFIG.md and
      doc/UI-FEATURES.md; the queue history path (per profile, not shared) and the
      Settings layout were corrected in the docs at the same time
    Downloads
  • v1.5.3-beta c130c4f458

    v1.5.3-beta
    All checks were successful
    build / check (push) Successful in 2m29s
    release / appimage (push) Successful in 4m39s
    release / rpm (push) Successful in 11m22s
    Pre-release

    Ghost released this 2026-09-12 00:15:40 +00:00 | 266 commits to master since this release

    Fixed in 1.5.3-beta

    • A second wizard run tried to re-create records that already existed: after a run the wizard dropped each written zone's cached records, so a later run against the same zone saw nothing cached, marked every RRset as new and failed on create; the post-run fetch now puts the fresh records back into the cache (and runs even without version history, where it used to be skipped entirely), so conflicts are detected again and Merge, Replace and Skip apply
    • Retrying a subdomain whose existence check had failed reported "Create zone failed": it now reports the failed check, because no create was attempted
    • The delete-zone reminder claimed too much: when the parent's records are not cached the app cannot know whether NS and DS records are still there, so it now asks you to check instead of telling you to delete them
    • Clicking an Override cell beside its checkbox cleared the row selection: which hid the Details box; the cell is selectable now

    Technical Improvements in 1.5.3-beta

    • The wizard's post-run queue item is called Refresh <zone> after wizard, since it refreshes the cache as well as taking a snapshot
    • The test fake for the cache now really drops a domain's records, so stale-cache regressions are caught; three new tests cover the refresh, including with version history off
    • Corrected stale defaults in the documentation: API rate limit 0.5 req/sec, sync interval 15 minutes
    Downloads
  • v1.5.2-beta 6f568ec5ec

    v1.5.2-beta
    All checks were successful
    build / check (push) Successful in 2m44s
    release / appimage (push) Successful in 5m12s
    release / rpm (push) Successful in 9m0s
    Pre-release

    Ghost released this 2026-09-11 22:19:16 +00:00 | 275 commits to master since this release

    Security in 1.5.2-beta

    • New API tokens' secrets were kept in the queue history: creating a token stored deSEC's response, including the token's one-time secret, as is: it was shown in the Queue page's detail panel and written in plain text to queue_history.json (readable only by your user). The queue now keeps a copy of each response with token values replaced by (redacted); only the token dialog gets the secret, once. History files written by earlier versions are scrubbed when the app starts, even with queue history persistence switched off. If a token created with an earlier version is still in use and your ~/.config/desecqt has been backed up or synced anywhere, consider replacing it

    Changed in 1.5.2-beta

    • README rewritten: install routes (AppImage, RPM, from source), use, rate limiting, DNSSEC and subdomains, privacy with what is kept on disk, documentation and support sections
    • Security reports go to support@jdneer.com: named in SECURITY.md and the README
    • Project links point at git.jdneer.com: the About page's project link and the changelog link opened GitHub

    Fixed in 1.5.2-beta

    • README license link: pointed to LICENSE; the file is LICENSE.txt

    Technical Improvements in 1.5.2-beta

    • scripts/test_api_queue.py (9 tests) runs in CI with the other headless unit tests
    Downloads