Re: Overbearing undownloaded face

[Kevin Greiner <kgreiner@xpediantsolutions.com>]

Harry Putnam <hgp@sbcglobal.net> writes:

> Simon Josefsson <jas@extundo.com> writes:
>
>> Hm. I must be missing some kind of use people make of the face.
>
> At one time I was a heavy user of the agent.  I can tell you that it
> can be confusing in a group that you read both on and off line.  I
> often had to back up and download a message that I had read on line
> but never downloaded.  Back then you couldn't tell.  It just showed
> read. (While online) So I can see some value in somekind of indicator.
> that is present even when online. What I take issue with is again
> related to reading both on and off line on the same server.

FYI, I am working on a group parameter that will provide fine-grain
control over which groups display the undownloaded faces.  Aside from
turning the faces on/off, the undownloaded faces will still function
as they do now.

> The undownloaded face server to show whether something is written to
> disk or not, but I think it should not over shadow other usefull
> marks one might want to use.  So, I wonder if there is isn't some
> other way to indicate downloadedness.  Not use a face at all.

Yes. Kai developed a format indicator at the same time that I worked
on the face solution.

> Kai says that is orthogonal thinking.  Running at right angles I guess
> he means, or talking apples and oranges.  But can't we use something
> in summary format line to indicate the downloadedness of an article
> and just drop the undownloaded faces?
>
> As Kevin pointed out that last part can be done by removing the 3
> default cons cells pertaining to downloadedness at
> gnus-summary-highlight.  That returns you to pre downloaded face
> behavior, which I favor.  But since I see usefullness in knowing
> downloadedness can't we do it another way that doesn't use an all
> powerfull face?

Now, you're coming closer to the problem.  What you're seeing as
arbitrary choices on my part are actually the result of constraints
imposed by the existing framework.

The summary format line, by virtue of being a string, makes it trivial
to extend its capabilities.  The downside is that 1) if I changed the
default, I'd annoy everyone as even unagentized summary buffers would
show a new column, and 2) those people who had already customized the
format line would have to read the manual to discover the new
feature. So, we have a new format but you have to go looking for it to
take advantage of it.

The face selection is handled by gnus-summary-highlight.  It's an
alist of conditions and face symbols.  It functions just like the cond
function.  The first non-nil condition will select the face.  This
structure is rarely customized as you're specifying both the face and
the precidence of faces.

In my opinion, this mechanism needs to be rewritten.  One of my first
releases of the undownloaded face was just that, one face.  A number
of people were upset that undownloaded articles hid the score faces so
I went back and created the high/low undownloade faces (3 faces
total).  Now, you would like to see a read/unread status (6 faces
total), ticked/unticked (12 faces total), etc. (Aren't permutations
fun :) ).  This is an option, but one that makes it difficult for
individuals to customize Gnus to a new color-scheme.

What we need is a highlight structure that specifies face attributes
(bold/italic/normal, red/green/blue, etc.) then matches on multiple
conditions.  Of course, we'd then need some sort of face engine to
find/create a face for each combination of attributes.  Anyone know of
one?

Kevin