I also tested a mail with mixed related/unrelated images, looks fine:

New issue for the issue of a partially unresponsive interface on mails with many/big mail attachments: T8356: GpgOL: Outlook might end up in an unresponsive state in read only mode for old mails

I talked with @mmontkowski:

• the description of the current state above is intended
• the display of the attachment, when the mail is opened the first time, can't be fixed
• the above mentioned issue of a partially unresponsive interface on mails with many/big mail attachments should still be checked, so i create another ticket for it

A pgp signed/encrypted message with drag&drop attachments (no CID i guess) will contain the attachments in an multipart/mixed part:

Looks good to me on gpg4win-5.1.0-beta658 / gpgol-2.7.3-beta11_dd853c7 @ win11:

Looks good to me on gpg4win-5.1.0-beta658 / gpgol-2.7.3-beta11_dd853c7 @ win11:

• The same errors are displayed
• Outlook does not freeze anymore

I assume, it's due to the difference in the mail structure (as displayed in outlook web):

Ok, I think, my tests above might be not valid, as new mails behave differently.
Some comment about what is to be expected would be nice, the last comment T7843: GpgOL: Empty OpenPGP mails with "Read as plain" activated is probably outdated.

I decided to add a retry count (=1). I check the implementation of usbd-ccid-0.4.0, it's ZLP (zero length packet) handling mistake.

Tested on gpg4win-5.1.0-beta658 / gpgol-2.7.3-beta11_dd853c7 @ win11.

Note: The fix will only apply to new mails, so I consider the old mails invalid now.

I'm retesting with newly created mails, maybe something changed

The original task should be tested with a VSD version but I did a quick check with Gpg4win-5.1.0-beta658 because of:

No change in gpg4win-5.1.0-beta658 / gpgol-2.7.3-beta11_dd853c7 @ win11:

With Gpg4win-5.1.0-beta658 Save secret team key is no longer offered in the file menu in the case of key on card:

In the keyboxd.log, the key is identified with e (ephemeral) flag and with no revoke flag (-).

2026-07-15 17:26:14 keyboxd[9452] DBG: chan_0x000000000000027c -> S PUBKEY_INFO 2 8D1B205094CF17E0ACB90DFBAA7CAA1F3303534E e- 0 0

@timegrid Thank you for the log. It helps me to identify the bug.
Here is the bug:
https://dev.gnupg.org/source/gnupg/browse/master/sm/keydb.c;0be940905d70125866622c6f53b8d25bde87c21b$1080?as=source&blame=off

Let me clarify "Won't fix". To change this, we need to modify the device and the protocol itself (and the functionality specification of OpenPGP card for the protocol). We "cannot fix" by scdaemon/gnupg. Background: It is common in smartcard culture that PUK cannot be tested alone, it can be only used with the update of new PIN together.

Looks good to me on gpg4win-5.1.0-beta658 @ win11.

Looks good to me on gpg4win-5.1.0-beta658 @ win11

• certificate details (context menu item added, tooltip enhanced)
  
• certificate list (tooltip enhanced)
  

Setting back to testing, to decide if this needs backporting

Looks good to me on gpg4win-5.1.0-beta658 @ win11:

In case it matters, those are the files i used above (signed by ted, created via kleopatra)

Without keyboxd, in both versions gpg4win-5.0.2 @ win11 and gpg4win-5.1.0-beta658 / gpg 2.5.21 @ win11, the import works (without the autoimport issue on search).

In T8347#220668, @werner wrote:

Which gnupg version is used by the gpg4win-5.1.0-beta658 Build. Please always mention the gnupg version if you use a gnupg26 tag.

Debugview output for gpg4win-5.0.2 (with auto import bug):

Which gnupg version is used by the gpg4win-5.1.0-beta658 Build. Please always mention the gnupg version if you use a gnupg26 tag.

Issue found on gpg4win-5.1.0-beta658 @ win11: