Graphs for average body?

[Peksa]

As a form of energy it'd be great to see avg. body graph. Specially effects of changing energy/body per cycle distribution rate and energy in corpses would be nice to see. Also someone (Nums? EricL?) spoke of graph for total energy, ie. nrg + body + shell etc, which would also be good to have.

[EricL]

As a form of energy it'd be great to see avg. body graph.

I'll put this on the list.   May be a month or two realistically.

Also someone (Nums? EricL?) spoke of graph for total energy, ie. nrg + body + shell etc, which would also be good to have.

Totday's total nrg graph does factor in both nrg and body (but not other things like shell) when calculating the total "nrg" a species has locked up in all it's members, but there is no real way today to see this on an average basis for organisms of a given species - jsut as a species total.  I'll put a "total resources individual average" graph on the list as well.  Same time frame.

[Peksa]

Thanks, I'm looking forward to it and other updates.

[Numsgil]

If you're feeling really ambitious it would be nice to allow people to construct graphs based on some set data.

Like: y axis = species nrg / species population.  Then people could mix and match for some really weird graphs.

[Peksa]

Oh, and imagine data being stored for whole sim history. Imagine seeing developement for hundreds of thousands or even millions of cycles instead of tens of thousands and still being to see graphs for last few thousand cycles if needed. Imagine being able to change something and see in the graphs' numbered x axis after days and days how it affected populations/energylevels/mutations in the long run.

Such a sweet dream.

[Numsgil]

Graphs are always sort of unrewarding to work on, but yeah, that's what I was thinking.  Would be so nice to be able to see long term what's going on in the sim.

[EricL]

I'm not very ambitious I'm afraid....  

I think the right way to do long term graphs or fancy post-processing of collected data is via the so called database logging feature using something like Excel or perhaps by enabling a graph itself to log its data to file.  Graphs arn't free from a perf perspective.  They take memory and cpu cyles to update.  Early on, I implemented code so that only those graphs currently being displayed get updated so as to not take the perf hit for all graphs all the time.  I don't think taking the perf hit for all sims for all graphs all the time just in case someone may want to look at historical info in the future is a good idea.  

The things I think are reasonable to do (meaning I'll probably implement them) are graph "features" (e.g. buttons to elect to show/not show zero valued series, change series color, etc.) additional graph types as requested, Internet mode enablment or Internet mode versions of current graphs where appropraite, mutliple graph instances of the same type that can run at different update intervals, more in-memory data points per graph for longer histories (say 1000 instead of the current 300) and decent long term logging to comma seperated format for off-line graphing and manipulation via excel or similar either via the current logging paradym or by logging enablment on a per graph basis.  I like that latter since it gives us a paradym for choosing what data we want to calculate and save (and what we don't) and at what interval.

The things I am unlikely to implement anytime soon include a graphing scripting language, super flexible runtime graphs with mix and match data types, that kind of thing.

[Peksa]

I'm not very ambitious I'm afraid....  

I was afraid that was the case

I think the right way to do long term graphs or fancy post-processing of collected data is via the so called database logging feature using something like Excel or perhaps by enabling a graph itself to log its data to file.  Graphs arn't free from a perf perspective.  They take memory and cpu cyles to update.  Early on, I implemented code so that only those graphs currently being displayed get updated so as to not take the perf hit for all graphs all the time.  I don't think taking the perf hit for all sims for all graphs all the time just in case someone may want to look at historical info in the future is a good idea.

I was afraid that that also would be the case. And probably savegame size.

The things I think are reasonable to do (meaning I'll probably implement them) are graph "features" (e.g. buttons to elect to show/not show zero valued series, change series color, etc.) additional graph types as requested, Internet mode enablment or Internet mode versions of current graphs where appropraite,...

These all seem to be basic but very useful features, especially being able to change almost white, imperceptible (woah, quite a word - I had to use a dictionary for that one) colour to something more useful.

...mutliple graph instances of the same type that can run at different update intervals, more in-memory data points per graph for longer histories (say 1000 instead of the current 300) and decent long term logging to comma seperated format for off-line graphing and manipulation via excel or similar either via the current logging paradym or by logging enablment on a per graph basis.  I like that latter since it gives us a paradym for choosing what data we want to calculate and save (and what we don't) and at what interval.

This would pretty much remove the need for continuous graphing. Being able to choose different update intervals for same data does exactly the same thing I was hoping for (even more if combined with ~1000 or more data points). It would allow graphs for long term and also for current events. Also, automated interval based logging for user defined data would be interesting. Everyone could construct the graphs they wanted to based on saved data. Hard drive space could be an issue here, but if it was optional it would be okay, especially when hard drives have become generally Huge.

The things I am unlikely to implement anytime soon include a graphing scripting language, super flexible runtime graphs with mix and match data types, that kind of thing.

Well those would anyway be something extra, like cacao powder on top of whipped cream on top of hot chocolate. Mmm... Not really necessary (though it'd taste damn good).


A couple of things you didn't mention but would be nice to have are sort of grid of values of graph on y axis and timeline for x axis and... Well, for now that'd be everything I can come up with graphs.