Fixed and backported for VSD 3.4.

Ranking as discussed with @ebo

The remaining issue "Wrong language in GnuPG error messages" is now also fixed. Until there's a new version of libgpg-error I've added the changes as patches.

The root cause is that opening the details reloads the certificate. This triggers a change of the key cache. And that triggers are reload of the group.

Backported for VSD 3.4

In T7495#209088, @timegrid wrote:

When an uncertified cert is uploaded, the main button of the warning message also says Export Certificate. Should this be another ticket?

This is now implemented for Gpg4win 5.

Likely a bug in Qt. Accessibility Insights for Windows shows that the parent of all grid items is the tree view. It should probably be the parent item instead so that ATs can go up and count the level.

In T6568#208992, @timegrid wrote:

Still open in this ticket (see https://dev.gnupg.org/T6568#208755 ):

• Collapsible items in tree views can't be expanded/collapsed, at least not with space or arrow left/right

Or Window activity (which is related to focus). I'm wondering if one of my patches is to blame.

I think Gpg4win will keep the language switch because for Gpg4win we don't disable any actions via action restrictions (as far as I know).

In T6568#208875, @timegrid wrote:

• Column sort order and visibility can be configured only via view and nowhere else. That's how it should be, right?

What do you mean?

The description says: The selection of visible columns and sorting should be possible **additionally** via the main menu.
I just wanted to make sure, "additionally" means additionally to the normal (non a11n accessible) context menu and not some other (a11n accessible) place i haven't found.

Yes, very likely it was with NVDA's (update) installer. Therefore I looked at NVDA's source code to find out how to make it work for Kleopatra just to find out that NVDA fakes it itself. I would close this ticket as won't fix.

In T7949#208870, @timegrid wrote:

Regarding the tests: Being able to set the language for Kleopatra via some environment variable would be fine, too.

In T6568#208750, @timegrid wrote:

Questions:

• Column sort order and visibility can be configured only via view and nowhere else. That's how it should be, right?

The 50 % is when either WKD lookup or keyserver lookup are done and the other is still running. None of those two lookups can be partially done. Hence there's no way to do something with more steps (safe for faking it with a timer). The only alternative is an unspecified progress bar, but unspecified progress bars (aka busy indicators) are completely silent as you can see in the "Loading certificate cache..." busy indicator. (NVDA's code doesn't support unspecified progress except in their own installer which has special code for this, i.e. the installer makes the sound itself.)

This isn't a Kleopatra issue. I suppose this happens with Outlook. The security confirmation checks for the type of keys it's asked for. Either OpenPGP or S/MIME or unspecified.

This isn't reproducible on Linux. (Tested with a Yubikey but that should really make no difference.) Maybe a regression of our Qt patches.

Please create a separate ticket for "The initial language in kleopatra is not selected according to windows settings". This is a bug in KDE Frameworks that's orthogonal to the issues of this ticket.

Best test this with a newer installer than gpg4win-5.0.0-beta413 to avoid the regression with the raw HTML (see T7886#208675).

The problem with the raw HTML has been fixed already by Tobias.

In T7759#208677, @timegrid wrote:

Forgot to note: Setting S/MIME debug level in kleopatra via GnuPG System will write the right key to gpg.conf (if I understood it right, this was also a problem)

You can also delete the `#include "deletejob.h" in job.cpp. Other than that it's ready for merge.

This has been changed:

• A progress/busy indicator page is now shown while smartcards are loaded (unless there are already smartcards shown; in this case the update happens "in the background"). Screen readers speak the progress message.
• The texts on the placeholder page (that is shown if no smartcards were found) are now accessible (if AT is active), i.e. one can reach them with Tab and initially the first text has keyboard focus (so that it's spoken).
• When a smartcard is found then the first smartcard tab has keyboard focus.

Except for a few minor things this looks good. Thanks!

Looks good.

In T7886#208534, @werner wrote:

FWIW, GnuPG uses this Windows API

[...]

but can be overridden with the LC_ALL, LC_MESSAGES, or LANG envvars by setting them to the usual xx_XX value.

This is fixed.

The error dialog now has a button to show the audit log (named Diagnostics).

The error dialog now has a button to show the audit log (named Diagnostics).

Please show the command line of the linker.

The revision was actually applied (rG0947a20c28cf: gpgsm: Fix output of card serial number in colon listing.), but Phabricator doesn't allow me to set it as applied because not all reviewers have approved it.

I think you are using an outdated working copy of gpgmeqt. I have killed the old jobPrivate stuff with rGPGMEQT056567525fb9: Add d-pointer to Job class. We are using Qt macros now.

Feel free to commit/push when you have made the suggested changes. One last thing: Update the NEWS file.

Werner is going to add the missing implementation so that the error will soon be gone.

In D618#6745, @TobiasFella wrote:

@ikloecker would there be any ABI problems with adding the new virtual function here?

Conclusion: gpg needs to emit a more useful status error. -> subticket

gpgme logs:

2025-11-13 11:22:26 gpgme[28014.6de1]   _gpgme_io_read: check: [GNUPG:] KEY_NOT_CREATED <LF>
2025-11-13 11:22:26 gpgme[28014.6de1]   _gpgme_io_read: check: [GNUPG:] FAILURE gpg-exit 33554433<LF>

where 33554433 means (GPG_ERR_SOURCE_GPG, GPG_ERR_GENERAL) = (GnuPG, General error)

For Kleopatra we need to add an "Audit log" button to the error dialog. And we need to check if gpg is giving us a useful error that we (gpgme) are ignoring or if gpg doesn't throw a useful error. What do the gpgme logs say?

Backported for VSD 3.4.