Re: existing work on TODO items

[Ken Manheimer]

On 1/8/06, Dave Love <address@hidden> wrote:
> Ken Manheimer <address@hidden> writes:
>
> > perhaps you're referring to my initial allout topic encryption
> > changes, which used facilities from crypt++ and mailcrypt?
>
> Yes, sorry.
>
> > someone suggested using pgg instead, and started some changes there
> > (to implement symmetric-key encryption) which would enable me to do
> > so.  i refined their pgg changes, added some general pgg fixes, and
> > switched over allout's new encryption to use the modified pgg.  it's
> > all checked in to the gnu repository, and comes with the head emacs
> > 22 checkout.
>
> I found some time to try, and it does run now, but I still can't see
> how it works (or, really, why).  I managed to get a section encrypted,

"why"?

i keep a daily log of activities, and am able to include details about
computer and financial accounts i create or change, including
sensitive things like passwords and lock combinations, inline,
encrypting them in subtopics of the activity.  likewise with sensitive
journal entries, and potentially i could encrypt topics targeted meant
for specific individuals or groups with their public keys to make
information available to them and only them, while being able to send
the files over public networks.  it makes it easy to use encryption
where and when i need it.

> how it works (or, really, why).  I managed to get a section encrypted,
> but it failed to decrypt, even when it didn't request a passphrase
> (cached from the original entry?), so I lost data.  Also setting

it would be helpful to me to find out how to repeat that bug, so i
could repair it.  (i'm assuming you're using the latest released
version.  i use the encryption feature frequently, and do not ever
have problems encrypting or decrypting.  i very much would like to
find out what went wrong in your situation.)

> `allout-passphrase-verifier-handling' didn't seem to allow me to use
> my private key.  There's no coding conversion done, so this can lose

it defaults to using symmetric-mode encoding.  describe-key (`C-h k
C-c x') will tell you that a single universal argument (ie, C-u) will
use key-pair encryption, and a doubled universal argument will do the
symmetric-mode encryption but disregard the cache.

> with non-ASCII, potentially also causing data loss, though I don't
> know what's actually done.

this is something i need to understand better.  it's a responsibility
of the pgg routines, but my changes may have left that off.  (if i
recall correctly, the coding conversion that was being done was
flawed, but i don't remember exactly.)

> > as far as i can see, outline.el provides only for navigating
> > outlines.
>
> I don't understand what you mean, and I doubt others will.  Outline at
> least does revealing of hidden items and subtree movement in some
> sense, though I don't know how they actually compare with allout and
> the manual doesn't document everything.  Exactly what behaviour do you
> mean?  Major modes should support Outline minor mode where reasonable,
> and a number do.

the standard outline mode has grown some features since last i
checked.  it didn't have any outline editing features before, only
navigation.  standard outline mode's addition of topic adjustment
(promotion, demotion, shifting) is useful, but
from cursory examination, allout still adds a number of significant
features.  off the top of my head, that includes:

 - topic/subtopic/supertopic creation by key-stroke
 - topic-oriented cut-and-paste across depths: pasting a
   cut topic into an empty topic header at a different depth
   adjusts the depth of the pasted topic, pasted subtopics,
   and pasted siblings
 - alternate topic header "bullets", to visually distinguish topics
 - ... including some programmatically significant bullets, like one that
   causes automatically maintained numbering of sibling topics ('#') or
   topic encryption ('~')
 - transient exposure of hidden search hits during incremental search,
   so they are reconcealed if the search continues past, or properly exposed
   if the search concludes there
 - key-stroke copying of exposed portions of a topic to different formats in
   a separate buffer, eg to indented or sectionally-numbered ("1.1.3.2") or
   latex formatting
 - topic encryption (which i am finding to be a slice-bread caliber feature, but
   it works for me)

> > the reverse is not the case - outline.el cannot handle most common
> > formatted outlines, much less custom ones,
>
> I don't know what that means but, for instance, my Python mode
> supports block-wise outlining with outline-minor-mode and Allout
> doesn't appear to support the outline variables that major modes set.

allout doesn't handle whitespace-delimited outlines.  (i also have an
unreleased minor-mode block-wise outlining package - outdent.el -
which i use for python programming.)  i don't know about the outline
variables that major modes set.  can you tell me about them?

> > outline.el lacks some of allout's navigation features (eg, hotspot
> > navigation),
>
> If I understand correctly, outline.el could easily do that, but I'm
> not convinced I'd want to be prevented from typing a letter into a
> Python block heading.  (outline.el should have had mouse-based tree

allout's hot-spot navigation is active only on the single
outline-header bullet characters - a reasonable analogue could be
somewhere in the leading whitespace on the first line of a block.  it
could be handy.

> manipulation like XEmacs', but as far as I remember it requires
> redisplay changes as for mouse-sensitive marginal icons, which I lost
> interest in pursuing.)
>
> There's clearly at least a documentation problem here if I'm confused
> even after a glance at the code and can't tell what it does for me.

i agree.  there are some provisions of which you might not be aware -
the allout-mode function has substantial documentation, as does the
primary encryption interface, allout-toggle-current-subtree-encryption
(the routine bound to `C-c x', mentioned above).

> Of course allout isn't unique in that, and I don't mean to criticize
> it particularly.  It definitely should address the potential data
> loss, though.

i agree.  can you give me more details, so i can fix it?

ken