[kplato] pert/cpm calculation (was Re: kplato)

[John Lamb]

Dag Andersen wrote:
> Ok, based on what we have discussed I've fleshed out some thoughts.
> 
> If we separate pert/cpm calculation code from the nodes but allow the 
> code to call the nodes methods, I think the following should be 
> possible. If you want to insulate the code more, we need some sort of  
> interface class. It will depend as much (more?) on your code as on 
> 'kplato' code, I think. What do you think?

It's probably too difficult to do the separation. Instead, I'll try some 
other way of isolating/testing stuff.

> (Note that pert_ is prepended for all methods that is only used by 
> pert/cpm calculation)

That's a really good idea, especially as it will work even within the 
KPTNode classes. What's important to me is to know what the algorithm 
may modify.

> I plan to allow resource overbooking here (for now?). Then the user 
> can resolve conflicts manually. 

That makes sense. Later, we can add code to help assign resources in 
complex cases.

> 
> I'm not sure where to eventually put more advanced resource analysis.
> 
> To get at the constraints of a node:
> node->constraint();
> If the constraint is ASAP or ALAP; no matter.
> If it's StartNotEarlier, FinishNotLater, MustStartOn or MustFinishOn 
> you need the time:
> node->constraintTime();
> 
> To set earliestStart/latestFinish:
> node->setEarliestStart(time);
> node->setLatestFinish(time);

Is this value set by the user or calculated by pert/cpm?

> To get a nodes list of child relations:
> node->dependChildNodes();
> 
> To get the relation type:
> relation->timingRelation();
> 
> To get the lag for a relation:
> relation->lag();
> 
> To set errors:
> node->setError(error);
> Type of 'error' to be defined...
> 
> To find out when to start/finish the node (within 
> earliestStart/latestFinish) and to calculate correct duration:
> node->pert_setStartFinish();
> 
> To get the type of node:
> node->type();
> which can be: Type_Project, Type_Subproject, Type_Task, 
> Type_Milestone, Type_Periodic, Type_TerminalNode, Type_Summarytask.
> Type_TerminalNode I think you will make disappear, Type_Subproject and 
> Type_Periodic are not implemented yet.

Type_Periodic will be the hard one!

> Type_Project, Type_Subproject, Type_Summarytask can have children.
> 
> To get a nodes list of children:
> node->childNodeIterator();

Looks good. I suggest you post what you need to be done. I'm happy just 
to write and test the algorithms.

-- 
JDL

Non enim propter gloriam, diuicias aut honores pugnamus set propter
libertatem solummodo quam Nemo bonus nisi simul cum vita amittit.