It does not say anything in the Release Notes about this, but the Strict Links functionality has improved in a way, that Spider gives an error, whenever he cannot respect strict links.
However on quite simple examples mentioned here and here Spider still cannot find a solution
So, I can see that work is ongoing in relation to Srict links, but I guess this requires quite a complex algorith to make it work properly.
Regards.
Member for
21 years 8 months
Member for21 years8 months
Submitted by Rafael Davila on Tue, 2013-11-05 14:55
RE: Not respecting a Strict Link without giving an error is in my view a defect (which, I think, is quite easy to fix)
Not respecting a Strict Link and producing some sort of "partial" solution is in error. I do not believe Vladimir will be comfortable with such a thing. Usually by further delaying both activities the link will be respected at some point in time while the resource requirements for both activities are also satisfied, obviously at some real cost but a solution different to partial solution of both activities. "Partial" solution means 2 activities wrong out of 2, a batting average of 0, it is no solution at all.
Based on your example I believe the functionality shall be improved and if it will not always enforce Strict Links maybe shall be renamed to something as Strong Links to mean the algorithm will attempt to enforce the link but will not always be able and in such case a flag will be issued. Then on schedule report the Strong Links not enforced shall be disclosed and a field shall identify non enforced Strong Links for the user to be able to filter.
Maybe instead of attempting for Strong Links a further enhancement would be attempting to Stong Maximal Lag, a Strong Link would be a special case of Stong Maximal Lag.
Best regards,
Rafael
Member for
21 years 8 months
Member for21 years8 months
Submitted by Rafael Davila on Mon, 2013-11-04 20:31
I never found strict links or resource leveling on hammocks reliable other than variable quantities and workloads with minimal set to 0, a valuable exception for many purposes.
Just change Activity 3 to a Hammock type with predecessor link to Activity 2 FS(0) and predecessor link to Activity 2 FF(2) resource loaded and make sure no link is of strict type. You will get similar behavior.
In both models create and display resource dependencies and the same wrong dependency will be created.
I do not mean some functionality for strict links cannot be programmed and perhaps in the form of a limited minimal/maximal lag functionality will be even better as a strict link is a special case of maximal/minimal lag functionality.
then, I think, we need to ask Vladimir to put this in his issues registry.
Not having optimal resource leveling can be considered as a tool just not being optimal, even though it is very unoptimal even in very simple cases (as in my 2nd example)
Not respecting a strict link without giving an error is in my view a defect (which, I think, is quite easy to fix)
Member for
24 years 8 monthsEvgeny,the work on improving
Evgeny,
the work on improving Spider Project is ongoing more than 20 years and will never stop.
You are right that strict links functionality is complex and shall be imroved carefully.
We know what shall be done, please wait. It will not take too long.
Vladimir
Member for
17 years 9 monthsJust tested the new
Just tested the new release 11.01.84 of Spider
It does not say anything in the Release Notes about this, but the Strict Links functionality has improved in a way, that Spider gives an error, whenever he cannot respect strict links.
However on quite simple examples mentioned here and here Spider still cannot find a solution
So, I can see that work is ongoing in relation to Srict links, but I guess this requires quite a complex algorith to make it work properly.
Regards.
Member for
21 years 8 monthsEvgeny,RE: Not respecting a
Evgeny,
RE: Not respecting a Strict Link without giving an error is in my view a defect (which, I think, is quite easy to fix)
Not respecting a Strict Link and producing some sort of "partial" solution is in error. I do not believe Vladimir will be comfortable with such a thing. Usually by further delaying both activities the link will be respected at some point in time while the resource requirements for both activities are also satisfied, obviously at some real cost but a solution different to partial solution of both activities. "Partial" solution means 2 activities wrong out of 2, a batting average of 0, it is no solution at all.
Based on your example I believe the functionality shall be improved and if it will not always enforce Strict Links maybe shall be renamed to something as Strong Links to mean the algorithm will attempt to enforce the link but will not always be able and in such case a flag will be issued. Then on schedule report the Strong Links not enforced shall be disclosed and a field shall identify non enforced Strong Links for the user to be able to filter.
Maybe instead of attempting for Strong Links a further enhancement would be attempting to Stong Maximal Lag, a Strong Link would be a special case of Stong Maximal Lag.
Best regards,
Rafael
Member for
21 years 8 monthsEvgeny,I never found strict
Evgeny,
I never found strict links or resource leveling on hammocks reliable other than variable quantities and workloads with minimal set to 0, a valuable exception for many purposes.
Just change Activity 3 to a Hammock type with predecessor link to Activity 2 FS(0) and predecessor link to Activity 2 FF(2) resource loaded and make sure no link is of strict type. You will get similar behavior.
In both models create and display resource dependencies and the same wrong dependency will be created.
I do not mean some functionality for strict links cannot be programmed and perhaps in the form of a limited minimal/maximal lag functionality will be even better as a strict link is a special case of maximal/minimal lag functionality.
Best Regards,
Rafael
Member for
17 years 9 monthsRafael,then, I think, we need
Rafael,
then, I think, we need to ask Vladimir to put this in his issues registry.
Not having optimal resource leveling can be considered as a tool just not being optimal, even though it is very unoptimal even in very simple cases (as in my 2nd example)
Not respecting a strict link without giving an error is in my view a defect (which, I think, is quite easy to fix)
Regards.
Evgeny