Re: [PATCH] Crypto-related (mostly) fixes and improvements

[Peter Bloomfield <PeterBloomfield bellsouth net>]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Albrecht,

On 08/10/2017 03:42:44 PM Thu, Albrecht Dreß wrote:
> Hi all,
>
> the attached patch fixes some issues with gpg/gpgsm encryption, and tries to improve encryption in general.  
> It also adds some other, small improvements.  Note that this patch includes the one I submitted a week ago 
> (subject “[PATCH] plug mem leak, simplifications”).
>
> (1) Bug Fixes
> - Due to an (undocumented) change in gpgme, S/MIME signing fails with error “not implemented” if the context 
> is created with a passphrase callback, so never set one in this case.
> - Fix several mem leaks in encryption.
> - Do not use a forced GnuPG key ID for S/MIME (see improvements, below)
>
> (2) Crypto-related Improvements
> - The user identity has a field for forcing the key id used when signing GnuPG messages.  Using it for S/MIME 
> is plain wrong, though (see Bugs above), so add an extra field for forcing a S/MIME certificate.  As to 
> simplify life for the user, add buttons to display the usual key list dialogue from which the user can choose.
> - Simplify the key list dialogue by showing the User ID only.  When the user double-clicks an entry, open a 
> new dialogue with the full key details.
> - Although exchanging GnuPG keys using a key server or (since the latest gpg versions) WKS is preferred, 
> there are situations where keys shall not be published (e.g. as to not disclose an employment of a person).  
> For this situation, add a send message option to attach the sender's public key.
> - On the receiving side, import application/pgp-keys parts into a temporary gpg context, and display the keys 
> with buttons for importing them into the main key ring.
> - Add subkey details (bits, type, ECC curve if applicable) to the key widget.
> - Make sure S/MIME signed messages always include the signer's signature (note that it does not make sense to 
> include the whole certificate chain, as the root cert should /always/ be imported from a trustworthy source).
> - RFC3156, sect. 3 requires a message to be 7-bit clean.  Thus, QP-encoding pure 7-bit parts is superfluous.
>
> (3) Misc Stuff
> - Simplify deleting a folder recursively and creating a temp folder by exclusively using glib functions 
> (re-sent from last week's patch).
> - Check BALSA_DIALOG_HEADERBAR for information dialogues.
>
> As always, any feedback would be highly appreciated!
>
> Cheers,
> Albrecht.

Thanks as always for the patch, and for the work it represents! Thanks also for rolling in last weeks 
patch--thanks to travel and an appalling connection, I hadn't even installed it :-(

The changes look good from a first look: Balsa builds and runs, but I've not tested all the code paths. I've 
pushed the patch to master, to get some serious testing beyond what I can do.

Best,

Peter
-----BEGIN PGP SIGNATURE-----

iF0EARECAB0WIQS030wPRfNNA5alz3MfX9S1uSp09QUCWY0OKQAKCRAfX9S1uSp0
9RMRAJ49YvgYXf1PIuoK5CXENpqxlUyZZgCcD7a9deO4DzJq2Zh6smNS7CK8+FY=
=nzny
-----END PGP SIGNATURE-----