Split GPG + Thunderbird + Enigmail produces empty attachments

[andrewdavidwong]

Conditions:

• Split GPG (qubes-gpg-split) 2.0.19
• Thunderbird 45.1.0 (probably also other versions)
• Enigmail 1.8.2 (probably also other versions)
• Inline PGP
• One or more attachments

Results:

• If signing only, a 0 byte detached .sig file is attached to the message along with the original attachment. (This should be a proper PGP detached signature file.)
• If signing and encrypting, the attachment is converted into a 0 byte .pgp file. (This should be a proper PGP encrypted message.)
• In both cases, there is no warning or error message.
• In both cases, the message body is still properly signed/encrypted.

[andrewdavidwong]

A workaround is to manually sign/encrypt all attachments from the command-line, attach them, then tell Enigmail not to sign/encrypt any of the attachments.

[andrewdavidwong]

A workaround is to manually sign/encrypt all attachments from the command-line, attach them, then tell Enigmail not to sign/encrypt any of the attachments.

[marmarek]

Cannot reproduce:

• when selected PGP/MIME, attachment is properly encrypted without any message,
• when selected inline PGP, TB asks what to do, with default to "Encrypt and sign each attachment separately and send the message using inline PGP"; after confirming, attachment is properly signed and encrypted

Tested on:

• Fedora 23 VM
• Thunderbird 45.1.1,
• Enigmail 1.9.4
• Split GPG 2.0.21

Maybe it was fixed as a side effect from other change? I'd guess adding support for --output option.
@andrewdavidwong can you confirm/deny?

[marmarek]

Cannot reproduce:

• when selected PGP/MIME, attachment is properly encrypted without any message,
• when selected inline PGP, TB asks what to do, with default to "Encrypt and sign each attachment separately and send the message using inline PGP"; after confirming, attachment is properly signed and encrypted

Tested on:

• Fedora 23 VM
• Thunderbird 45.1.1,
• Enigmail 1.9.4
• Split GPG 2.0.21

Maybe it was fixed as a side effect from other change? I'd guess adding support for --output option.
@andrewdavidwong can you confirm/deny?

[andrewdavidwong]

Indeed, it looks like it's been fixed as a byproduct of something else. 👍

[andrewdavidwong]

Indeed, it looks like it's been fixed as a byproduct of something else. 👍