Technical problems?

Descendant chart worked for me without any 'DNA' restriction; as did the Y-DNA for a male 'focus' profile. mtDNA requires a female 'focus' profile -- and that one works, too.

You will note the restriction I found earlier, the the maximum number 'graphable' seems to be on the order of 4000-5000.

When I'm not sure if it is a 'data problem' vs. a 'program problem', I try the graph with just 1 or 2 generations. If that "works", but it doesn't for (say) 6 or 7 generations, then it is more likely to be a graphing / data limitation of graphics package, not a failure of the application to access the Geni data.

Dan, I was thinking I may put a limit in HistoryLink in the descendant listing so that if it has goes above 4000 for example, it stops and perhaps indicates to the user that they try again at a lower point in the tree. Do you have a suggestion on the # at which it might want to do so or should I just leave it as is?

There are a couple of alternatives & factors that have come to mind:

1) I suspect that the issue with graphing has to do with width of the smallest 'sector' being calculated -- maybe that would be something you could estimate and thus 'stop' before the graphics just 'hang up' (that would depend on being able to get a rough value for the size of the sector and the number of descendants from that sector and calculating an approximate resulting pixel size and stopping if, say < 3) In other words, checking the graphing limits because the graphing package doesn't do so.

2) Whether one estimates the 'graphing limit' or just uses a descendant count, the real question may be: What to do at that point?

a) Gracefully stop.
b) Ask the user.

Choices for (b) might be:

i. Stop & display current incomplete graph.
ii. Stop.
iii. Continue counting descendants, but do not attempt to graph.
iv. If 'DNA limited', skip graphing, but list living descendants. (Come to think of it: If not DNA limited, identify Geni member profiles so one might contact them.)

Leaving 'as is' is a bit of frustration because it doesn't give a lot of guidance one what one might do next.

Just 'thinking out loud'...

Thanks Dan, for engaging as well. It feels better if I know other technical minds are on call for reports about Jeff's app as well. The SA users love this app, but it means I feel like I'm constantly whining to Jeff for free help. (If I can whine to you too - it spreads the load :-))

The graph is building again for my own ancestry, but still sticking on the mtDNA descendants of others. (Which – I could be mistaken, but – I think are a lot fewer in number?)

Eg 1) A previous mtDNA search: http://historylink.herokuapp.com/graph?profile=profile-g60000000004... on Inabe (Ansela) van Timor, SM/PROG produced: Loulette Laubscher & Sterna (Roux) Schreuder as claimed Geni profile descendants.
But no longer?

2) http://historylink.herokuapp.com/graph?profile=profile-g60000000048... for the mtDNA descendants of Francoise Martinet, SM/PROG stops after 5 generations, but we know that at least, Private User is definitely one of her mtDNA descendants with a claimed profile on Geni.

We know this from genetic tests, and also by tracking the matrilineal line, profile by profile, on June’s Geni family tree view.

(However, if you prompt Geni to find a relationship connection – by clicking on the green pin – it finds a non matrilineal line between the two: [ ie June Barnes → Catharina Wilhelmina Buchanan (Jansen van Rensburg) her mother → Stephanus Petrus Jansen van Rensburg, g1h7 her father → Maria Jacomina Jansen van Rensburg, f2g3 his mother → Catharina Johanna Susanna Jansen van Rensburg (Nel) her mother → Gerrit Nel, d3e5 her father → Jan Nel, b2c2d3 his father → Jan Hendrik Nel, b2c2 his father → Susanna Fourie, b1 his mother → Suzanne Cordier, b1 her mother → Francoise Martinet, SM/PROG her mother]
Is this perhaps the cause of the History link tree build breaking off after 5 generations?

3) And the mtDNA graph http://historylink.herokuapp.com/graph?profile=profile-g60000000062... for Sarah Barentsz Cochet, SM/PROG will not build at all – even for 2 generations; and even for ancestors, (Although this may be due to someone having succeeded in adding a living profile Private User as her daughter in law: which I’ve reported twice to Geni)

Sharon: re: (1) Those two profiles (Loulette Laubscher & Sterna (Roux) Schreuder) are 11generations away from Inabe (Ansela) van Timor, SM/PROG. Using the relationship 'stick pin' to find the path, and then starting with Ansela van de Caap, SM/PROG actually identifies 3 claimed profile descendants (1 of them male).

(2) There are 4 generations of males (3 in a row!) between June & Francoise, so she won't show up on the mtDNA chart.

I tried showing All descendants of Francoise, and then visually looked at the female lineage. When I got to Martha Susanna Groenewald as the sole female with further female descendants -- her two daughters don't have any descendants at all listed on Geni. So that seems to be why that graph -- for mtDNA -- seems to stop: because Francoise apparently does not have direct female descendants after that ... based on the current Geni connections.

(3) Is definitely a Geni problem (not Ancestor Graph's). The Tree View does not display correctly, and attempting to look at the Relationships tab also fails.

Thanks Dan, that's a lot of help. (And yes, of course that's how I did No 1 - How did I forget that??)

Re No 2: if you actually go mother by mother by mother on June's tree view - it ends up at Francoise Martinet. I'm thinking that she must have two lines to Francoise (& Geni is just preferencing the one including males) - but I can't figure out why the historylink graph should be stumped by that.

re #2. If I do an 'Ancestor Graph' from June, the 10th generation does indeed get to one of Francoise's daughters (Suzanne Cordier, SM).

Then, starting with that Suzanne, a Descendant Graph (10 generations) does indeed find June (+ 5 others).

-- That explains why June would not show up in Francoise's 10-generation descendant (she's 11 away).
-- Does not really explain why it seems to "stop" graphing at 4th generations of descendants from Francoise; it should find June's mother!

Jeff: That 'Descendants' of Francoise just quits graphing after generation 4 ... even though it indicates it is searching through generations 5->10.

Sharon Doubell - I'll take a look at the Francoise issue. It looks like it actually does the research, but fails to display it. If you run it on Francoise's daughter Suzanne, all the descendants list and June is displayed (for those of us that have permissions to see that deep). So one of the other daughters of Francoise must be causing an issue in the graphing function. I'll have some time tonight that I can work on it.

Thanks Dan - edit conflict. :)

Sharon Doubell & Dan Cornett - Problem should be fixed. The issue was Agatha Dorothea Retief, who has double quotes within the country name ("La Motte", Drakenstein, South Africa). The double quotes were escaping the JSON data string used to create the graph. I've now added a check to replace all double quotes " with a single quote ' in the location fields for graph generation. I also put in a stop if the descendant graph goes above 2000 after a generation renders.

Actually, I'm going to remove the 2000 check until I can do more testing on it. I'm thinking it would be better perhaps to do something like more than 1000 in a single generation (ring). Or maybe I can stop it from graphing such a ring when zoomed out but allow it when viewing a pieces zoomed in. Try to give the user some options...

Jeff ,
Can you raise the number of generations to 12 or would this just increase the problems.
10 generations just does not reach to living persons from 1650
regards
Peter

Peter ... I suspect once the graphing issues are resolved, then the limitations on the maximum number of generations checked can be better dealt with.

Jeff: Great idea about limitations on the number per ring! If you just 'suspend' the graphing when a ring gets "too dense", maybe just keep the total update going (along with the 'researching' at the top) until completed.

Good to know about double quotes & JSON ... do you already check that on names?

Thanks Geni Magician

Dan, yes - I was already removing the double quotes on names as some people include things like Marie "Mary" Smith, but I wasn't checking locations fields - oversight.

Peter Dennis...Gone fishing!, I've raised the max number of descendants that can be researched to 14. However, I placed a limit on the number of descendants in the outer ring HistoryLink will display before it stops. If a generation count (ring) filtered by DNA exceeds a set # individuals (currently 1000), it will stop researching. The graph is not currently capable of effectively displaying more. So if your person from 1650 does not have an overly dense descendant tree on the selected search (e.g. mtDNA, yDNA, or All) and it's just deep, then you should hit those living individuals. So we'll see how this works... If it's both dense and deep, then I don't yet have a solution.

As Dan stated, I'm dealing with some limitations on the graph (and processing that many descendants on the backend). Figuring out how to dynamically extend the graph to show denser areas when zooming has been proving difficult. I'd like to allow it to keep the graph running and continue the count, but currently I count as I iterate the individuals in the graph. So I'll have to look at how to do it outside this process, but I also don't want to count 15,000 descendants, so I'll have to find some type of balance and the user may just need to graph lower in the tree.

... I'm o.k. with just 'stopping the graph' when it gets too dense. It would be nice to have the details 'pop up' when selecting a sector (profile) in an outer ring, but having to go through re-setting the focus person is (in my view) a minor inconvenience. (Besides, that can be done on a separate tab/window.)

Here's another thought ... maybe ... allow one to reverse the 'logic'/direction on the original HistoryLink, adding options to show Geni-member profiles (of descendants). Then you don't have to worry about graphing at all; just list those "identified" by the selected options (e.g. projects, Geni-members, problem profiles, ...).

You might need to still limit the number of generations in that direction (e.g.: someone attempting to start with the biblical Moses or Abraham), but that might be another alternative for finding mtDNA and Y-DNA descendants.

Jeff: You may want to look at the code a bit more about the increased # of generations.
There is a rather sparse descendant tree starting with Ebanezer Greaves -- that only graphs 10 'rings'.
Starting down 4 'rings' with Thomas Greaves (1624), I get a graph of 10 'rings'.
Starting down 7 'rings' with Thomas Greaves (1732), I get a graph which finally shows active Geni members and has (only) 9 'rings'. (Probably no more descendant generations stopped this one.)

If the search works correctly, I would assume that the 'Geni members' ought to have been found when starting from the 1624-Thomas.

(Options were All [gender] Descendants for 14 generations ... supposedly.)

Well, I was finally able to get Geni to work in Firefox! I found an old thread from 2007 on a Mozilla help forum about a similar log-in problem and someone suggested deleting the cookies from Geni. Once I did that and reloaded the page, I could log right in. Plus, History Link works great now, including the fantastic graph!

I had one question though -- is it possible to set the graph to be figured from someone on my tree other than myself? I have only been able to add my father's side to my tree, so it makes more sense to figure from him, rather than having my mother's half blank.

Yay on the "delete cookies" trick working. Cookies get hairy if left out too long.:)

Yes, very easy to change the ancestor graph focus person

Go to http://historylink.herokuapp.com/graph

Ancestor Graph for Erica Isabel Howton, (c)

There's a little person icon to the right - click it and you'll see

Change Person

Put the URL or the Geni profile ID into that box, press the refresh (circle icon) to ensure it took and the profile came up correctly.

Proceed with the rest of the parameters you want to set:

Generations to Display: of:   DNA:
Build Graph
Save Graph

Thank you, Erica! Worked perfectly! It really shows me where all the holes are though -- I've got a lot of work to do.

Private User - in addition to my 'report' on May 16th, I found another 'minor' issue -- which, perhaps, everyone ought to be aware of:

I tried running two different browser tabs, on two different profiles (which I don't think have any profiles in common), doing full Descendant Graphs -- at that same time. One graph displayed as expected ... the other did not (started o.k., but didn't finish o.k.) I re-established the focus person on the second one, ran it by itself, and it then displayed fine.

I've only tried that once. (Firefox on Windows.)

(tech note: probably some shared global variable that got confused)

So, until that is fixed (if ever) ... only run one Descendant Graph at a time (at least, for one browser).

Thanks Dan, ya - I think I'm temporarily storing data based on your userid, so when you kick off a second session, it can overwrite the temp data of the prior search. I probably need to shift it over to the profileid being searched, so each has it's own temp. I need to look at memory management.

re: May 16th findings: that 'oddness' hasn't changed.

Ah, must have missed that - will look at it today.

Dan Cornett, Thanks again for the report and the follow up as I overlooked it on May 16th. I've fixed the issue that was limiting it to 10 generations.

I had put in place data verification in the backend code as to prevent someone from changing the ancestor count to something higher than I allowed (since it's a post function) as not to blow up the app, but after I updated the front end code to support 14 generations of descendants, I never updated the backend code to allow for more than 10. That has now been fixed (and I increased it to 15 generations). Also fixed a bug where it was giving "UK" and "United Kingdom" different country colors.

I haven't yet gotten to the multiple session bug.

Great about the fix!

FYI: I'd prefer first the Private/Living/Deceased feature before the multiple session "convenience".

re: latest 'fix' on descendants

Seems the name labels are a bit broken. Try that Ebanezer Greaves descendants for more than 10, with labels 'All'.

... re: labels ... only does 10 (toggle between '9' and 'All' shows that)

I noticed that while you were typing and already fixed. :)

Dan Cornett, could you repeat or link to the "Private/Living/Deceased" request. I plan to work on HistoryLink a bit this weekend. I'd like to add a feature to the HistorySearch that will allow you to Follow all the profiles in your ancestry search. I'd like to set mine to 15 generations and hit go to make sure I'm following all my ancestors.