Problem using GenSmarts with TMG
Posted May 5, 2007 4:46pmMatt_G writes...
Here is the issue that I just noticed.
With certain individuals, if I invoke GenSmarts from within TMG using Ctrl+E or the menu, I do not get any research suggestions for that person. However, if I run GenSmarts as a standalone program and put that SAME person's ID number in the search box on the To Do tab like such [6451] I get numerous suggestions. All the suggestions I get ARE for the person with that ID number as well. In other words I am not confusing suggestions which may apply to another principal of an event like a wife on a marriage tag.
What is strange is that this is only a problem with some people. With others it works just fine.
This is a big issue for me because I want to use the ability to create research tasks from the GS suggestions that WhollyGenes coded in.
This ability isn't available when running GS in standalone mode. It can only be done when GS is invoked from within TMG.
Any ideas?
Matt
On 2007-05-05 7:31pm GenSmarts replied...
In stand alone mode, GenSmarts is analyzing your entire TMG file, whereas when called from TMG, it's only working with what TMG is passing. Pilot error aside (using different files, an old verison of GenSmarts or TMG, etc.), I would think the issue lies in the data that TMG is picking out for GenSmarts to base the analysis on. You can see this data (it's in XML format, which is pretty readable) in the TMG program directory after you make the request to GenSmarts. I don't remember the file name exactly, but it should be pretty obvious. There'll be one file used as input, and one file that comes back with the output.
I would take a look at the data being passed by TMG and see if you spot anything missing... then ponder if you have some option set somewhere in TMG that might be causing that data not to be passed. Obviously, it could be some kind of bug in GenSmarts or TMG, but this area is pretty shaken out by now and I don't recall any known issues related to what you're describing. There is a bug or two in TMG related to ampersands and other non-alpha characters, but I'm pretty sure that causes an exception/error, not just a lacking of results.
On 2007-05-05 10:39pm Matt_G added...
After looking at the XML files I may have found a problem. I'm not sure.
I invoked GS from inside TMG on a person where everything works.
In gsdata1.xml I see the following at the beginning of the file:
<GEDCOM>
<!-- Family Records -->
<FamilyRec Id="FM1">
<HusbFath>
<Link Target="IndividualRec" Ref="IN701" />
</HusbFath>
- <WifeMoth>
<Link Target="IndividualRec" Ref="IN700" />
</WifeMoth>
IN701 is Display ID #14 which is the person who had focus when GS was invoked. IN700 is Display ID# 13 which is his wife. Like I said, for this person everything works perfectly.
Here is what gsdata1.xml looks like on his daughter where it does NOT work.
<GEDCOM>
<!-- Family Records -->
<FamilyRec Id="FM1">
<HusbFath>
<Link Target="IndividualRec" Ref="IN688" />
</HusbFath>
- <WifeMoth>
<Link Target="IndividualRec" Ref="IN689" />
</WifeMoth>
- <Child>
<Link Target="IndividualRec" Ref="IN700" />
</Child>
</FamilyRec>
<FamilyRec Id="FM2">
<HusbFath>
<Link Target="IndividualRec" Ref="IN701" />
</HusbFath>
- <WifeMoth>
<Link Target="IndividualRec" Ref="IN700" />
</WifeMoth>
- <Child>
<Link Target="IndividualRec" Ref="IN13143" />
</Child>
- <Child>
<Link Target="IndividualRec" Ref="IN13144" />
</Child>
- <Child>
<Link Target="IndividualRec" Ref="IN13146" />
</Child>
- <Child>
<Link Target="IndividualRec" Ref="IN13148" />
</Child>
- <Child>
<Link Target="IndividualRec" Ref="IN13149" />
</Child>
- <Child>
<Link Target="IndividualRec" Ref="IN13151" />
</Child>
- <Child>
<Link Target="IndividualRec" Ref="IN13152" />
</Child>
</FamilyRec>
IN688 is Display ID#1 which is her GRANDFATHER. IN689 is her Grandmother. She is married. Shouldn't she be in this entry of the file?
She is only in the file as her father's child under the FM2 section. The person I was trying to get suggestions on is IN13144. Her husband is IN13145 and isn't listed anywhere in the FM1, FM2, or FM3 sections of the xml file.
Is this the problem?
If so, how the heck do I fix it?
I hope I have made sense here. If not I will be happy to clarify.
On 2007-05-06 1:32am GenSmarts replied...
In the gscontrol.xml file, is TMG asking for suggestions for the correct person? In other words, gsdata1.xml is simply the context for to analyze in making suggestions. Given your description, I would expect you'd find this entry:
<Suggest Target="IndividualRec" Ref="IN13144"/>
I wouldn't think data on a single grandparent would have that much bearing on this - it would only be used in trying to estimate things that were missing on the primary individual, and you wouldn't think you'd have to go all the way to a grandparent to round out a birth, death, or marriage estimate. On the other hand, if you have really sparse data about this person and her immediate family, perhaps it might have an impact.
However, a husband is pretty useful in coming up with estimates as they tend to live in the same time period and often the same places... so I could see where a missing husband would impact the suggestions. Check how you've defined the marriage to TMG - anything unusual?
If it turns out incomplete data is the culprit, you'll have to run your example by TMG support and see if they can give you any more info on what TMG should be writing to that file and if it's simply a TMG bug that it doesn't do that, or if it's some setting, preference, or corruption in your data that's causing the problem.
As GenSmarts uses the same algorithm for both standalone and "invoked by TMG" mode, the suggestions will be exactly the same - if the input data is the same. TMG could decide to filter some of those suggestions, for example, if TMG sees that you already have added it to your TMG To Do list, but I don't think that is the case currently (TMG doesn't filter anything that comes in from GenSmarts).
Unrelated to TMG, but a "no suggestion" problem in standalone mode is typically an inability to make sense of the place names in the data. Applying that to your problem... you might double check the place names in the XML for anything unusual compared to the examples that are working for you. I recall an example from a year or so ago where somebody had some undesirable XML file affects from using a special TMG "don't print prefix" on some of their place data...
