Re: Re[2]: Idea: Dependencies between Tasks
Mikael Hallendal <[email protected]>
| Newsgroups | gmane.comp.gnome.apps.mr-project.user |
|---|---|
| Message-ID | <[email protected]> |
mån 2002-07-01 klockan 22.42 skrev Michele Ravani: > On Mon, 01 Jul 2002 13:00:03 +0100 Ian Phillips <[email protected]> wrote: > > IP> Hi Michele, > IP> > IP> Maybe I'm missing something, but I don't see how your "events" are any > IP> different > IP> from breaking a task down into sub-tasks. > > Hi > > There is no difference, it is an alternative which I think would add > flexibility and provide a 'light-weight' way to define dependencies between > tasks. > > Assume that you have a task worth 10 days of development effort which is > being done by a single developer, Joe. He knows what to do and could take the > task and run with it. In a way, such a task is a 'natural' work unit. > > However, another developer needs for his upcoming work, the API of the > module in development, which is going to be ready (if all goes well) after > 3 days worth of work. > > I find that having to split the task in two (or more) just to be able to > build my dependencies correctly is cumbersome: create a new task, describe > it (not really necessary, Joe knows exactly what he has to do), assign a > resource (with the chance that resource leveling will put it somewhere in > 2010), an effort, etc. > > An 'event' would simply be a satisfied condition or a state within a task. > I would have to give it a label, may be a planned completion time (date or > % of task) and eventually set the flag. > > In principle, these 'events' are nothing else but a way to define > dependencies between any two points within tasks. > > I often met situations as the example above in the past, where one had to > decide between the bore and effort to define zillions of almost senseless > tasks (ok, ok I am exagerating a bit), or accept that the dependency > network was in principle incorrect. In the example you described there _is_ to phases (or more) of the large task that you have. The design of the API is it's own subtask and should be described as such. I think that having some kind of "events" that does the exact same job as the subtasks (in a much fuzzier way) is only confusing and doesn't really solve anything that isn't done correctly by using subtasks. Regards, Mikael Hallendal -- Mikael Hallendal [email protected] CodeFactory AB http://www.codefactory.se/ Office: +46 (0)8 587 583 05 Cell: +46 (0)709 718 918