Re: new problem with Balsa (git head)

[Jack <ostroffjh users sourceforge net>]

Hello Albrecht,

On 2018.02.01 14:48, Albrecht Dreß wrote:
> Am 01.02.18 20:15 schrieb(en) Jack:
> > I actually just had the problem again, this time while fetching  
> > mail, and even the POP3 fetch dialog stopped "spinning" until Balsa  
> > started responding again.  Whatever the cause, it looks like it's  
> > interrupting/blocking the program at a fairly deep level.  Also -  
> > every message that is selected when this happens does show the PGP  
> > lock.  I also think the reason it is so sporadic is that only a few  
> > of my mailboxes (for IT related mailing lists) have any messages  
> > which are PGP signed.
>
> In order to trace the timing of the GpgME operations, you can enable  
> debugging as described here:  
> <https://www.gnupg.org/documentation/manuals/gpgme/Debugging.html>.
>
> I also noticed that Crypto recently tends to be somewhat slow, in  
> particular if the local key ring and/or the TOFU db is large.  I  
> don't know, though, if this is related to gpgme or gpg.  The balsa  
> code interfacing to gpgme hasn't been changed since ages.  A quick  
> comparision between using gpg2 from the command line and through  
> gpgme seems to be somewhat slower for the latter, but I don't have  
> real numbers yet.
>
> Please check if the gpg option “auto-key-retrieve” is enabled, as it  
> may also slow down the operation.
>
> As a quick workaround, you might want to change the decrypt/signature  
> check option for the critical mailbox to “never”.  In this case, the  
> crypto backend is called only if you explicitly request the operation.
>
> I will look into this issue…
Thanks.  Changing to "never" does seem to have worked.  Am I correct  
this is a per mailbox setting, with no general override?  Might be a  
nice enhancement.

To repeat, my problem is not how long the check takes, but that it  
blocks Balsa from doing ANYTHING at all.  However, I can imagine the  
potential issues if you (for example) go flying through (down arrow) a  
whole bunch of messages, and have to track (or cancel) those checks on  
the fly.  I'm actually trying to do some real work right now, but will  
check the debug output within a few days.

Jack