Comments about Finding and Reminding

Yesterday on IRC, Allan Day kindly gave me some feedback about the
Finding and Reminding prototypes that I posted a while ago [1].  

With his permission, I'll reply here in the list, to keep it referenced.

> (03:21:48 PM) aday: federico: so about your finding and reminding
> thingamy
> (03:22:21 PM) aday: some random points...
> (03:22:43 PM) aday: * it feels very different from the existing views
> in the overview - i want more parity

Hmm, do we need it to look similar?  It's showing a different kind of
information - Windows is for "what windows do I have open?",
Applications is for "what apps can I run?", and the journal is for "what
stuff have I done recently?".  Even Windows and Applications look pretty
different, have different ways of interacting with them, etc.

> (03:23:55 PM) aday: * i'm not convinced by the 'pending' part on the
> right - i don't want to be reminded of what i have to do every time i
> go to open a document, nor do i particularly like the idea of
> introducing more 'management' work that a user has to do to allocate
> documents there
> (03:24:14 PM) aday: this thing should require as little user 
> management as possible, preferably none

Yeah, I'm not entirely happy with the reminders section yet - all the
"Pending" areas.

I did it that way to borrow some concepts from Getting Things Done, in
particular the tickler file, while trying to make it less "formal" or
with a less-rigid way of using it.

I want an informal, but reliable way of scheduling things - "I need to
deal with this sometime next week" - not something as rigid as, say,
Evolution's to-do list.  Or maybe we just need an informal-looking way
to use Evo's infrastructure as it is; it would certainly be good to
explore that (e.g. does VTODO support DTSTART/DTEND or DTSTART/DURATION
properties, and can we use that to represent "sometime next week"?).

Suggestions for how to schedule things are very much appreciated.  I'm
not happy about the clutter that the reminders section adds to the
journal view.

> (03:25:20 PM) aday: * why have everything/new items only/frequent?
> again, there's a lot of work for the user to do - can we not just have
> one order that works?

Oh, we can play with those to see what is really worth keeping.
Originally I just wanted "Everything" ordered by date:

	Thu Jun 02
		blah.txt foo.jpg eek.doc

	Wed Jun 01
		eek.doc bar.odt trollcat.png

Then, Jon suggested that one's memory for past dates isn't really that
accurate, and thus having specific days is not very useful, so I divided
the time buckets to have a finer granularity for recent dates, and
coarser for older dates:

	This week
	Last week
	This month	
	Last month
	Last month - 1

Seif wants to have "smarter" modes as well, like the frequently-used
items, and filetype-specific items ("show me only Pictures, sorted by

I do agree that the fewer options we have for showing the journal, the
better.  We just don't know which options will be the best ones yet.
> (03:26:41 PM) aday: * finally - and this is the big issue that mccann
> was referring to - non-local data has to be a top level priority. that
> means we have to figure out a story around what it means to 'save' and
> what our open dialog looks like

We had a big argument about this the other day, but I'll summarize my

People who need to Get Work Done on computers (as opposed to hipstering
in Facebook and Flickr all day), at least in my part of the world, still
use plenty of local apps.  They do word processing, they make
spreadsheets.  Graphic designers photoshop all day.  They have many
little problems when dealing with pushing their files from app to app,
or when having to arrange their daily, ephemeral work.  These people
move files around on tangible, physical USB sticks, and they don't know
or care about cloud services like Amazon S3.

I want to help those people; I think it's easy to do, and it doesn't
require writing a lot of infrastructure.  This is outlined in (I need to update that page a
bit, but it's still accurate).

I *do* want to have non-local data show up usefully in the journal as

	- What documents did I frob in Google Docs?
	- What git pushes did I do?
	- What photos did I upload?
	- What Dropbox things did my brother share with me?
	- What interesting web pages did I read?
	- etc.

But that requires machinery that we don't have yet.  We may need to
scrape the HTML of some web apps to figure out what the user did -
imagine some GreaseMonkey code that figures out, "you edited this doc in
Google Docs" and pushes out that information to the Zeitgeist log so
that it will show up in the journal.  We may need a way to identify
things that you have stored in S3, Dropbox, or whatever.

What I *don't* want to do is spend a year or two figuring out how to
interact with services that I honestly don't use very much.  I think the
time-based journal is a good foundation to maintain your everyday
working set, regardless of remote/local stuff, and *later* we can see
how to make non-local apps work with it.  We have to do this

> (03:27:10 PM) aday: so it looks like there are other tasks that need
> to be dealt with before finding and reminding can be properly dealt
> with - which is what mccann is doing, i beleive
> (03:27:19 PM) aday: phew, so that's about it :)

Thanks for the feedback!  And if you think of something better for the
reminders, I'm very open to suggestions.



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