[Spambayes] Web UI features request: multi-select on the Reviewpage

Meyer, Tony T.A.Meyer at massey.ac.nz
Sat Jul 26 11:09:25 EDT 2003


> I mean that the Web UI should operate as a valid W3C 
> compliant HTML form.

In what way is the UI not w3c compliant in terms of forms?  (There are
lots of things that do need to be fixed for it to be generally
compliant).

> >  * You can definitely select more than one item at a time 
> > (in fact you can't select only one).
> Of course you can select only one:

I meant "select" as in "have selected at the time the form is
submitted", not "select" as in "change with one click".

> the form opens with all of the items in 
> the Review in their relevant pre-assigned categories and asks you to 
> either accept all of those assignations, or go through and 
> individually alter each selection [mail item] by altering them
[clicking 
> on them] one at a time. 

Or to change them all (to defer, for example), and individually alter
any that you wish to, one at a time.

> There is no mechanism to [on a PC] allow for the use of Ctrl Click or 
> Shift-Ctrl Click to select sets of user selections from within the 
> Spambayes assigned sets and then reset all those selected to 
> a different category e.g. setting several of those currently set as
ham 
> to spam or vice versa as a single operation.

The standard operation of [Shift-]Ctrl-Click is to select an individual
item for later attention.  It would make no sense, as far as I can see,
to individually select messages so that you can then click on 'Spam'
(for example), when you could just click on the spam button for each
one.  If you are ctrl-clicking on 10 messages, for example, you then
have 11 clicks (one for each message, and one for the classification).
With the current method, you have 10 clicks.

Standard shift-clicking, of course, is different, since here you click
on boundaries, and have the items within the boundaries also selected.
So, assuming that the 10 messages were sequential, you could make do
with three clicks.  This is a big assumption, however, since ham & spam
don't arrive (unless by chance or location) in sequential groups.

> That is not a valid interpretation of multi-select under the 
> W3C HTML 4 standards. A multi-select is precisely what it says: a
users' 
> multiple selection from within the existing categories, not an all or
nothing 
> acceptance/rejection/alteration of all of them. 

What standards are you reading?  This, right?
<http://www.w3.org/TR/html4/interact/forms.html>

There is nothing in there (as far as I can see) about "multi-select", or
multiple selection from within existing categories.

> In simpler english Spambayes presents messages incoming 
> cached messages under three possible category headings
> [spam, ham, defer].

I'm afraid that your simpler English doesn't make sense.  Do you mean
"presents cached messages", or "presents incoming and cached messages",
perhaps?

> >  * Plus it defaults to training as whatever it was 
> > classified as, so you only need to change those that it got wrong,
> > or defer/discard those that you don't want to train for some reason.
> 
> Precisely: with several hundred messages, needing to select 
> several dozens of messages individually from within a few hundred
preset 
> items is a royal pain.

1.  If "several dozens" of messages are being incorrectly classified
(out of "several hundred"), then spambayes is not working well for you,
and you either need to do more training, or fix something that's not
correctly set.

2.  You have to signify somehow which messages need to have their
classification changed.  With a control-click, you will have to click
each one individually.  A shift-click might work, but would require
incorrectly classified messages to be sequential (in sequences of four
at a minimum, for any effort to be saved).

I'm sorry, but I still don't see what the improvement you are asking for
is (which obviously means that I can't add it).  Can you perhaps point
me to an example form somewhere that demonstrates the feature you are
asking for?  Or a specific part of the w3c recommendation that the form
is not complying with?

=Tony Meyer



More information about the Spambayes mailing list