NetAccess::statInternal and showing job progress

Aaron Peterson myusualnickname at
Mon Jul 19 09:31:23 BST 2010

I often want to disconnect from a network as soon as files are done
(Think Mission Impossible :)       So I want to at least be able to
actively check notification log to find messages such as these. I
think everyone is annoyed by tons of pop up notifications.

Here are two ways I want to be able to wrangle my notifications.

implement a timer that checks a condition before doing the
notification and action. This would be an optional extra parameter
sent along with the notification....  And this would help solve other
problems too!  bug 215229

implement a log viewer / analyzer plasmoid.

1 Log analyzer,
IM style Notifcation log display
2 DelayTillAction
and relation to
3 Rambles


Do we not have an instant messenger style logging mechanism?  I looked
thru the plasmoids for one but did not see one.
----We should implement a log analyzer in one. ( Think server grouping
failed log in attempts )

Failed login(x5)

We need an algorithm to  determine if a notification is worthy of our attention:

Information (somewhat) available / that could be sent by an
application to help our log viewer choose what to show...  ( also
would help us implement a free facebook alternative with our instant
messangers... !!! We could trigger retransmissions of some events if
log analyzer rule is met...  But I digress)

1. expected quantity minimum- and maximum-in the next timeframe.

2. Chain/Thread ( a program may send many events, but many could be
related, simply coming from the same program might not be enough, but
a good start would be each instance of an application uses a unique

3. Delay between event and consequence
Typical bandwidth for LAN,
Typical bandwidth for WAN
Typical bandwidth for Local
Time before battery runs out

4. Severity,  -- an application may try 3 times complain quietly,
before issuing a notification when it tries a 4th time..


   Condition to check
   Time to check  (or delay till check)
   Action(s) to take
   What to do if cannot check condition

This would also be a fix for this bug:

We could calculate a quantity of time for an event to trigger if we
knew bandwidth, size/bandwidth*1.33    We could estimate 10mbit/sec
for internet, and 50 for lan?


 when expecting email, I'd like to know that kmail is talking with the
server. basically positive notification is what I want, but of course
I might start ignoring it, and would like "negative" notification if
it is failing. -------   Basically, a log viewer ---- that I can check
----that takes constant space --- that does not steal focus, is all I
can hope for.

People back in the day kept an xterm open in the corner to view
notification of events.  They could scroll, it took constant space,
and it didn't steal focus.

Users could use their log analyzer to tweak notification severity.

 It appears that we use dbus/knotify for critical errors (network
outages, power supply failure...)  So applications may also need to /
be able to report on severity, and re-transmit a critical level
warning instead of waiting for knotify to reply.

On Fri, Jul 16, 2010 at 12:45 PM, Vitor Boschi <vitorboschi at> wrote:
> Hiding notifications when opening small remote files would also be nice. I
> find unuseful (and even somewhat annoying) to show them when the download
> time is so small that you didn't even see the progress bar.
> Vitor
> On Fri, Jul 16, 2010 at 4:37 PM, Chani <chanika at> wrote:
>> On July 16, 2010 11:42:48 Aaron J. Seigo wrote:
>> > hi ...
>> >
>> > NetAccess::statInternal is a private method and is used by calls to the
>> > NetAccess methods for stat and exists. if the url is remote, it shows
>> > job
>> > progress. this means that for every file that is stat'd or checked for
>> > existene over the network (e.g. fish), we get a job notification.
>> >
>> > this results in apps like ksnapshot, when used to save to a remote
>> > location, throwing up job notifications on start up.
>> >
>> > it also means causing a lot of traffic in the job UI server
>> > infrastructure
>> > just to show what is, imho, fairly uninteresting information to the
>> > user.
>> >
>> > i have a simple patch here which changes this:
>> >
>> +1000
>> *hugs*
>> yaaay, less unuseful notifications! :)
>> (note: anyone who makes ksnapshot not try and access remote locations on
>> startup also gets hugs. it really wouldn't hurt to wait until I click save
>> to
>> autogenerate a filename.)

More information about the kde-core-devel mailing list