emacs-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: gpg-agent support removed?!


From: Ken Manheimer
Subject: Re: gpg-agent support removed?!
Date: Mon, 10 Apr 2006 18:40:36 -0400

daiki, i still have some problems with your patch, and the changes
you're suggesting for allout, and i think you've oversimplified the
problems allout has having with your patches.  details interspersed
with the cited text.

On 4/8/06, Daiki Ueno <address@hidden> wrote:
> >>>>> In <address@hidden>
> >>>>>   Romain Francoise <address@hidden> wrote:
> > Richard Stallman <address@hidden> writes:
>
> > > Romain, would you please explain to me why you removed the code in
> > > Emacs CVS?
>
> > To keep it level with the same code in Gnus, which we planned to include
> > in a new Gnus stable release this week-end.
>
> > > Why not install Daiki's fixes instead?
>
> > Because they weren't available at the time,
>
> As I said yesterday, the current version is at
> http://www.unixuser.org/~ueno/pgg-20060407.tar.gz
> This code should be fairly stable.
>
> > and because we had agreed to test this new code in the Gnus trunk.
>
> I agreeded to revert the code from v5-10 branch of Gnus because the
> release of Gnus v5.10.8 is so imminent.  However, I didn't think the
> release of Emacs is as well.
>
> > Anyway, Reiner has now changed it again in Gnus and Emacs to a later
> > version that satisfies everyone, I think (revision 1.8 vs. 1.4).
>
> No, I wasn't satisfied with your decision at all.  Though allout.el
> appearantly has a bug, i.e. allout.el uses PGG in a bad way, it works as

you characterized it that way because allout uses some internal pgg
functions.  that was not complete, nor did your suggested remedy solve
that problem, though it would allow the removal of those functions
from pgg-gpg.

first, your initial patch crippled allout's passphrase provisions
because the pgg-gpg functions which took an optional passphrase
argument were not respecting it.  i mentioned that, and you seem to
have fixed it - that problem no longer exists with your more recent
patch.  that's very important to allout's encryption operations, and
i'm glad it's resolved - thanks!

second, you suggest fixing some of the cases where allout uses
internal functions by having allout implement its own version of those
functions, allowing you to eliminate the versions of those functions
in pgg-gpg.  that mostly works, but it amounts to allout's versions of
those functions depending on internal data structures in pgg-gpg,
leading to a yet more brittle situation than depending on internal
functions.  i think a better fix would be to provide the
functionality, or some equivalent, at the generic pgg level.  i did
not do so because i had limited time and no access to the other
backends, in order to establish that the functionality would be
available with them.

it also so happens that there is at least one other function from
pgg-gpg which allout depends on and which is missing from your revised
pgg-gpg: `pgg-gpg-lookup-all-secret-keys'.  this one calls out to pgg,
and so does not even remotely belong in allout.  perhaps it was
renamed or the functionality is available in another way, and i'm not
opposed to tracking that, but i cannot do so immediately, and would
need time and justification to do so.

> expected by Ken (the author of allout.el), so it won't be fixed.  On the

now i'm confused.  what gave you the idea that it won't be fixed?  i
never said i wouldn't consider what you suggested, and in fact have
been planning to do so but haven't had the time until now.  i think
the approach is flawed, for the reason i mentioned above, but you may
be able to convince me otherwise - i'm willing to talk about it.

> other hand, I have many enhancement plans of PGG based on the new code
> such as keysigning features, string-based interface, etc.  I think it
> should be tested not only by the Gnus people but also by Emacs people.

i think pgg could stand some attention, and am glad you're willing to
do so.  the thing is, taking it may require some consolidation before
extending it - it has grown a bit unordered, and probably needs some
attention to rationalizing it against existing consumer's
requirements, including those of allout.

for now, i have a branch version of allout that does as you suggested
and works partly against your pgg-20060407.tar.gz patch, except that
key-pair encryption fails for lack of pgg-gpg-lookup-all-secret-keys,
as mentioned above.  i am going to be going away for a week, starting
tomorrow evening, so i'm hoping that there will not be a migration to
your new version until this is resolved.  i ultimately would prefer
that the problems be resolved in a better way, along the lines
mentioned above, but am willing to go with a partial solution if some
provision for its stability can be made.  i do want to see pgg
continue to ripen.
--
ken manheimer
address@hidden
http://myriadicity.net




reply via email to

[Prev in Thread] Current Thread [Next in Thread]