Now that I've changed my noise-detection-configuration, wp and wr timers run at 
about equal speed for restbreaks and micropauses (see bug 30).

Daily limits, however, is a different story. There is still a factor two 
difference. Most likely, this can be attributed to the fact that wp needs 30 
seconds to detect that I'm idle (i.e. the daily limit counter keeps running for 
an additional 30 seconds)

I'm not sure what to make of this. I could, of course, reduce my limit to 2 
hours, to compensate for the measuring differences. There are, however some 
drawbacks to this approach
- I cannot explain my employer that I only work two hours a day
- After reaching daily limit, I feel like having worked a lot. I cannot explain 
to myself that it's only been two hours.

The alternative would be to mimick wp behaviour, introducing different idle 
times for different timers. This, too, has it's drawbacks
- Timers are no longer consistent
- Configuration more difficult

Like I said before, I don't know what to make of this. Any thoughts??

Noise detections per timer is overkill. Perhaps, simply disabling noise
detection for the daily limit timer will do? Rationale: if you generate what is
currently seen as noise, you're sitting behind the computer. Your boss would see
that as you working...

Don't think that would do. During a day's work, I generate relatively 
little "noise", due to the rather strict settings.

However, If I become active, wp subtracts at least 30 seconds from my daily 
limit. wr currently only does 5. I think this much better explains the 
difference between wp and wr.

Anyway, from an architectural point of view, there is not much difference 
between introducing per-timer idle times, or per-timer noise detection.

Just making everything configurable is not really the answer, IMHO. We really
need to think what we want and supply good defaults whenever possible. Not just
throw in an option or emulate WP just because ...

So, what do we want? 

> "I cannot explain my employer that I only work two hours a day"

So, is our goal to have the daily timer match the time you spend behind the
computer (which is what your employer sees you doing), even if you are not at
all very active.  That's why I at first suggested disabling noise detection for
the daily timer.

Now, I propose the following:

daily timer idle IFF micropause timer has reached its autoreset and is idle.
in all other cases

(daily timer is active otherwise)

There went something wrong with the final conclusion, here goes again:

daily timer idle IFF micropause timer has reached its autoreset and is idle.
(daily timer is active otherwise)

Btw, rationale:
- your employer (and you) probably view micro-pauses as a part of your working
time. Therefore, you working time should only halt if after a (possibly natural)
micropause you still haven't started working.

Eek... you're doing it again... asking me what I want. We've been there... I 
want a guarantee that wr will warn me "on time". (whatever that may be).

Anyway, answering this question has to do with what your users in general will 
want. There are a few items I can think of:

- They want to feel good. They are (probably) not going to use a program that 
makes them unhappy. For RS-Injured people, feeling good is roughly proportional 
to (amongst other things) being able to work a significant amount of time. In 
this sense, the only requirement is that the timers represent a "fair" measure 
of the amount of work that is done, and they need not be correct to the last 
millisecond. There still is debate possible on the definition of "fair" (which 
is why I filed this bug). But after a days work, 4 hours feels fairer than only 
two.

- They want to compare themselves to others. One of the most frequent questions 
heard amongst injured people are "how are you" (duh) and "how much work do you 
do". Initially, this is the amount of time spent in the office (when working 
parttime), later this is the amount of time spent behind the computer. These 
people are not helped when every rsi-program measures time differently. 
Therefore, I believe we should not deviate too much from the industry standard 
(wp), even though our approach is "more correct". (Like I said before "correct" 
is not a requirement :-)

As for me, I consider myself assimilated. I have a daily limit of (only) two 
hours now.

Ok. A new configuration options is available:

/timers/<timer_name>/monitor = <ref_timer_name>

Semantics: timer 'timer_name' will be idle IFF timer 'ref_timer_name' is 
idle and has reached its auto-reset limit.

regedit is your friend (or gconf-editor)

[HKEY_CURRENT_USER\Software\Workrave\timers\daily_limit]
"monitor"="micro_pause"

Initially, at startup, a timer has not yet reached its auto-reset limit, so the
daily limit timer will active at startup... Perhaps we have to set the idle time
of all timers to auto-reset at startup ?

Now also configurable through preferences.