Update to Notifications Wiki Spec

6 messages Options
Embed this post
Permalink
Mimi Yin

Update to Notifications Wiki Spec

Reply Threaded More More options
Print post
Permalink
Some javascript/style in this post has been disabled (why?)
Hi Brian,

I've updated the wiki spec for notifications with the following changes:

1. More style guidelines for how to display notifications
2. Notifications are now more specific, with potential to expand in-place edits made to the notes field
3. Reports re: upcoming events and ticklers are in a separate section
4. Affordance for expanding the list of notifications you see
5. New time zone label next to username to show what time zone the widget is operating in

New open issue:
Can we differentiate between when someone manually triages an item versus when it automatically pops to NOW because of a ticker alarm or the event start date/time rolls around?

Brian Kirsch

Re: Update to Notifications Wiki Spec

Reply Threaded More More options
Print post
Permalink
Some javascript/style in this post has been disabled (why?)
Thanks Mimi,
See comments in-line.


On Mar 13, 2008, at 12:48 PM, Mimi Yin wrote:

Hi Brian,

I've updated the wiki spec for notifications with the following changes:

1. More style guidelines for how to display notifications
2. Notifications are now more specific, with potential to expand in-place edits made to the notes field
3. Reports re: upcoming events and ticklers are in a separate section
4. Affordance for expanding the list of notifications you see
5. New time zone label next to username to show what time zone the widget is operating in

New open issue:
Can we differentiate between when someone manually triages an item versus when it automatically pops to NOW because of a ticker alarm or the event start date/time rolls around?


Hum, I don't have the answer at the moment. Randy would be the best source for this question. However, based on my research
I would guess this is not possible at the moment.

Perhaps with the addition of the Event / Notification log Database table the code could guess that the triage change was
due to an alarm or start date time by looking at those values.  



Randy Letness

Re: Re: Update to Notifications Wiki Spec

Reply Threaded More More options
Print post
Permalink
Brian Kirsch wrote:
>
>> *New open issue:*
>> Can we differentiate between when someone manually triages an item
>> versus when it automatically pops to NOW because of a ticker alarm or
>> the event start date/time rolls around?
>>
>
> Hum, I don't have the answer at the moment. Randy would be the best
> source for this question. However, based on my research

We can't really tell the difference.  The best we could do is make a
"guess" by looking at the last modified date and comparing that to the
event start date or reminder time.  If the last modified date is before
these times then we can assume the item was manually triaged, otherwise
assume it popped to NOW automatically. Also,  if the "autoTriage" bit on
an item is set to false, we can assume the item was manually triaged.

There are a couple of limitations on the sever that we have to deal
with.  First, the server can't determine if an item was triaged as NOW,
it just knows the item changed.  So if the item was already triaged as
NOW and the item changes again (say the description was updated), there
is no way to determine that just the description was updated.  So there
is no way for the server to say "only triageStatus changed".   Also,
items never change on their own on the server.  In order for an item to
change, a client has to sync those changes.

-Randy

Mimi Yin

Re: Re: Update to Notifications Wiki Spec

Reply Threaded More More options
Print post
Permalink
Some javascript/style in this post has been disabled (why?)
Randy and I chatted briefly, here's my summary:

1. Phase 1: With relatively little work, we can get a Notifications widget up and running that just know *that a change or changes* were made to an item, *Who* made the last change on an item, *When* it was made.
- This means that if a Desktop client syncs with a whole mess of changes to a single item made over the course of a day of being offline, all those changes will get lumped into a single change record: Lars changed xxx at zzz time.

2. Phase 2: With a lot more server work, the server can keep a log of *what changes were made* on an item over time:
- Lars changed the Title of xxx to yyy at zzz-1 time
- Lars changed the Notes of xxx to yyy at zzz-2 time
- Lars changed the Notes of xxx to yyy at zzz-3 time

* The server would be able to track changes at the granularity of "per edit session" or "per set of changes committed to the server when user clicks [Save] button"

* The server won't have this kind of information from syncing Desktop clients. Instead, all edits will be lumped together with the last edit time: 
- Lars changed the Title of xxx to yyy at zzz-3 time
- Lars changed the Notes of xxx to yyy at zzz-3 time
- Lars changed the Notes of xxx to yyy at zzz-3 time

3. Phase 3: With additional work on the Desktop client, we can keep a log on the Desktop client for every edit to every attribute over time, just like the server in Phase 2.

Both web and desktop client will eventually need to understand the difference between users manually assigning triage status to items and Chandler auto-triaging items to NOW based on event start date/times and *shared* alarm date/times.

Randy, does this sound right to you?

Mimi

On Mar 17, 2008, at 8:14 AM, Randy Letness wrote:

Brian Kirsch wrote:

*New open issue:*
Can we differentiate between when someone manually triages an item versus when it automatically pops to NOW because of a ticker alarm or the event start date/time rolls around?


Hum, I don't have the answer at the moment. Randy would be the best source for this question. However, based on my research

We can't really tell the difference.  The best we could do is make a "guess" by looking at the last modified date and comparing that to the event start date or reminder time.  If the last modified date is before these times then we can assume the item was manually triaged, otherwise assume it popped to NOW automatically. Also,  if the "autoTriage" bit on an item is set to false, we can assume the item was manually triaged.

There are a couple of limitations on the sever that we have to deal with.  First, the server can't determine if an item was triaged as NOW, it just knows the item changed.  So if the item was already triaged as NOW and the item changes again (say the description was updated), there is no way to determine that just the description was updated.  So there is no way for the server to say "only triageStatus changed".   Also, items never change on their own on the server.  In order for an item to change, a client has to sync those changes.

-Randy
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Open Source Applications Foundation "chandler-dev" mailing list

Randy Letness

Re: Re: Update to Notifications Wiki Spec

Reply Threaded More More options
Print post
Permalink
Mimi Yin wrote:

>
> *3. Phase 3:* With additional work on the Desktop client, we can keep
> a log on the Desktop client for every edit to every attribute over
> time, just like the server in Phase 2.
>
> Both web and desktop client will eventually need to understand the
> difference between users manually assigning triage status to items and
> Chandler auto-triaging items to NOW based on event start date/times
> and *shared* alarm date/times.
>
> Randy, does this sound right to you?

Yes, with the third phase still being a little fuzzy.

-Randy


Graham Perrin

History, playback, timeline, presentation of log of edits to attributes (Was: Update to Notifications Wiki Spec)

Reply Threaded More More options
Print post
Permalink
At
<http://lists.osafoundation.org/pipermail/chandler-dev/2008-March/009782.
html> (referred from
<https://bugzilla.osafoundation.org/show_bug.cgi?id=11936>), Randy
Letness wrote:

> Mimi Yin wrote:

>> Š 2. Phase 2: With a lot more server work, the server can keep a
>> log of *what changes were made* on an item over time:
>> - Lars changed the Title of xxx to yyy at zzz-1 time
>> - Lars changed the Notes of xxx to yyy at zzz-2 time
>> - Lars changed the Notes of xxx to yyy at zzz-3 time
>>
>> * The server would be able to track changes at the granularity of
>> "per edit session" or "per set of changes committed to the server
>> when user clicks [Save] button"
>>
>> * The server won't have this kind of information from syncing
>> Desktop clients. Instead, all edits will be lumped together with
>> the last edit time:
>> - Lars changed the Title of xxx to yyy at zzz-3 time
>> - Lars changed the Notes of xxx to yyy at zzz-3 time
>> - Lars changed the Notes of xxx to yyy at zzz-3 time

>> *3. Phase 3:* With additional work on the Desktop client, we can
>> keep a log on the Desktop client for every edit to every
>> attribute over time, just like the server in Phase 2.
>>
>> Both web and desktop client will eventually need to understand
>> the difference between users manually assigning triage status to
>> items and Chandler auto-triaging items to NOW based on event
>> start date/times and *shared* alarm date/times.
>>
>> Randy, does this sound right to you?
>
> Yes, with the third phase still being a little fuzzy.
>
> -Randy
Does Google Wave change any thinking in these areas?

The 'Playback' feature appears at around 13:10 in the Google Wave
Preview movie.

A little discussion:
<http://chandlerproject.org/script/getIrcTranscript.cgi?messagesOnly=true
&channel=chandler&date=20090531>

Regards
Graham

_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev