Re: Agenda clock reporting

[Carsten Dominik <carsten.dominik@gmail.com>]

On Jan 3, 2010, at 4:38 PM, Bernt Hansen wrote:

> Carsten Dominik <carsten.dominik@gmail.com> writes:
>
>> On Jan 3, 2010, at 3:22 PM, Bernt Hansen wrote:
>>
>>> Carsten Dominik <carsten.dominik@gmail.com> writes:
>>>
>>>> I think the right place would be `org-clock-sum'.  Right now,
>>>> this looks only for complete clock lines.  You could modify the
>>>> regexp
>>>> to check for incomplete lines (first time stamp only) and then  
>>>> check
>>>> if the org-clock-marker points to the current line.  If yes, you
>>>> could
>>>> use the current time as the end time and proceed.
>>>>
>>>> Send me a patch if you get this to work, with a customization
>>>> variable
>>>> to turn this off.
>>>>
>>>> HTH
>>>
>>> Yes it does thanks.  I found the logic that uses the regexp for
>>> parsing
>>> the times and summing them in reverse up to the headline in
>>> org-clock-sum.  I haven't modified the regexp yet but was planning
>>> something similar to what you describe above.
>>>
>>>  - org-clock-sum
>>>  - Runs through headings in reverse order
>>>  - counts clock entries bottom up to headline
>>>  - adds a text property to the headline with total time
>>>  - need to find out if current heading is clocking
>>
>> Yes, the alternative approach would be to *not* modify the regexp,  
>> but
>> to
>> check if org-clock-hd-marker points to the current headline and then
>> add the time since the clock started (in `org-clock-start-time') to
>> the appropriate
>> field in the ltimes vector before extracting the total time.  So you
>> would do
>> this as the first thing when you find a headline, and the proceed
>> normally.
>>
>> Sounds simpler than modifying the regexp and would not cause any
>> problems
>> with other incomplete clock entries that may have been left behind  
>> and
>> never
>> properly resolved......
>
> That does sound simpler ... I'll give that a try.

In fact, you only need to add the time to t1, not to the ltimes  
vector.....

- Carsten