Family Historian User Group

I've noticed since downloading the upgrade to FH6 that when I type in a place that it auto fills by default to American places whereas most of my places are in the UK-is there a way of changing the default?

There have been several comments about how Place (and Address and Occupation) fields auto-complete, and recent changes to it.

The general opinion is that it used to be based on the most recent or most often used data, but now is based on some alphabetical preference, and several users are not happy with it. However, it is not biassed towards USA or UK data.

There are no settings you can change, but just keep typing the Place name ignoring any auto-completion until you get the Place you want.

You can complain to Calico Pie via http://www.family-historian.co.uk/suppo ... t-overview and the + Contact Support link.

So I'm not the first! I've contacted Support at Calico Pie so perhaps they'll change it next time round, he said hopefully!

Well I hope so too. Its a great time saver when entering whole families to just be able to type the first letter of a Place when using it in consecutive entries.
Anne

Had a reply from Calico Pie regarding above, they said:

We're puzzled by this. FH does take into account your current location, but it gets its information about your location from Windows. Is it possible that Windows thinks your PC is an American PC?

In any case, you can solve the problem within FH. Click on the Tools menu, then Preferences, and on the Map Window tab, set the 'Override Default Country Code' to 'GB'.

I've not got an American pc so that's weird but hopefully the solution will help everyone else?

John, I think the question you posed has been misunderstood.

The answer suggests they think you are talking about auto-geocoding Lat/Longitude in the Map Window, rather than auto-completion of Place names in the Property Box.

Perhaps you could post here a copy your question you sent to Calico Pie.

I can't find the original that I sent but I have the reply and have sent a 2nd reply quoting your comment, Mike.

Had another reply from Martin at Calico Pie as follows:

OK sorry. But there is no country-bias in auto-filling with place names. What you are seeing must be just chance.

So what can we do about that?

See my original reply to you on Fri Mar 25, 2016 11:40 am that explains the options.

That is a very poor response from Martin at Calico Pie.
Clearly, the Place auto-completion process is not some random chance, but a programmed process based on previous Place field entries.

I suggest you contact Calico Pie again, now they understand the area of concern, and ask how the Place auto-completion process works now compared to earlier versions of FH. Ask if our perception correct that it is now 'alphabetic' rather than 'most used' or 'most recently used'? Ask them to explain how the algorithm works now and how it worked in previous versions.

OK I will try again using your words.

Had a reply from Martin as follows:

As far I know, the auto-completion processing has not been touched since it was first written. Yes you may have found changes in the way it works in practice, between versions, but if so, these are a knock-on consequence of other changes and not a result of changes to this feature as such. I can't tell you how the algorithm works. A quick test suggests that in any given session, it initially matches alphabetically, and thereafter matches on the most recently-used value. But that may not be correct.

We have logged the fact that you're not happy about the way it works now, and we will look look into possible improvements in the future.

So I guess wait for next upgrade?

I've logged a message to Martin saying that this issue was clearly raised during FH6 beta, in which context his reply is, IMHO, distinctly inadequate.

I get the impression that Calico Pie have either hired some new staff, or perhaps temporary staff while the senior members are at the WDYTYA exhibition.
I recently got a reply that advised I should spend some time getting more familair with FH, and had to remind them who I was and that I had in depth knowledge of FH from Version 2 onwards.

Ouch! Did you explode, or just burst out laughing? They must be VERY new staff if they have never heard of you.

I have invested a little time reviewing this scenario.

Firstly, FH does NOT save any Place name usage details other than what is in the Gedcom file.
So, the bias to most recently used Place names is only compiled during one FH session, and lost on closure.
This agrees with my experiments in both FH V5 and FH V6, and the recent reply from Calico Pie.

I have compared the FH V5 with the FH V6 copy of the Family Historian Sample Project.
In both there is an entry beginning with Bristol that is used more often than an entry beginning with Bath.
In FH V5 when you initially enter B it auto-completes with the Bristol entry, i.e. most used.
In FH V6 when you initially enter B it auto-completes with the Bath entry, i.e. alphabetic.
The same pattern repeats with Place names starting with C and with L.

I suspect that in FH V5 it is counting how many Place fields use each name.

I suspect that in FH V6 it is simply looking in the Place records and setting each usage count to 1, even though Tools > Work with Data > Places does have a Used count the same as FH V5.

So the auto-complete algorithm has definitely changed, and I will report my findings to Calico Pie.

Hi Mike

This feature appeared in version 3.1.2 :

If you start to enter a place name into a place field, auto-completion will now choose the most-recently-used matching place name as the 'auto-complete' version, for preference, if there is a choice of more than one place name that matches the typed text. If the text does not match the most-recently-used place name, the most-frequently-used matching place name will be chosen, if there is a choice of more than one. The same applies to auto-completion in all other contexts (e.g. religions, etc).

I remembered this because my Gedcom Census (now Ancestral Sources) software chose the 'most common' place name and I'd made this suggestion to Calico. So I was pleased they did this but they trumped me by using the most-recently-used first, an even better solution .

But I agree that in v6 this no longer works correctly.

Cheers

Nick

That is a useful quote Nick.

The most-recently-used matching place name feature still works in FH V6, but the quote omits to say it only works within one session, so at the start of a session it does not apply.

The most-frequently-used matching place name feature is what is broken in FH V6, and I suspect it is not the auto-completion algorithm itself that has changed, but the way the Place name list is compiled with the same usage count for every name.

Note that the quote includes: "The same applies to auto-completion in all other contexts (e.g. religions, etc)."

I just tried it with religion and it seems to work correctly there: when file loads up it works on most frequently-used rather than alphabetical order. So I think you're correct, the algorithm still works the same way but it is the usage counts for places that is broken. Probably related to the work done for v6 with Place records.

Nick said

I just tried it with religion and it seems to work correctly there: when file loads up it works on most frequently-used rather than alphabetical order. So I think you're correct, the algorithm still works the same way but it is the usage counts for places that is broken. Probably related to the work done for v6 with Place records.

That was exactly the point I made to CP when I wrote to them last Friday. Maybe they'll now give the problem a little more attention.

I think I disagree that in a new session, Place names are alphabetical. Every time I try to enter 'Stepney', it autofills with 'Stepney Green'. If it was aphabetical, Stepney would come before Stepney Green.
This example is particularly annoying, as I have far more entries using Stepney, than I do for Stepney Green!

A <Space> character is before a <Comma> alphabetically, this results in what you are seeing.

I must admit, I never thought of that, thanks for the explanation. I'll now ensure I enter an additional space before the comma wherever it causes a problem like this.

EDIT
Just changed 'Stepney,Middlesex,England' to 'Stepney ,Middlesex,England' and problem resolved, it now appears before 'Stepney Green,Middlesex,England', alphabetically!
Thanks again

Mike, in earlier versions of FH you would not have that problem, because the more-frequently-used Stepney would take precedence over Stepney Green, so you would not need the workaround. That is what we are all complaining about, but Calico Pie are not listening. Perhaps you could also Email them about your problem, and with weight of numbers they might fix it.

Another oddity is that the alphabetic order of such entries is different in the Place Records Window compared to the Tools > Work with Data > Places window.

BTW: Some tools such as Queries and Plugins may not correctly match Place parts with a trailing space, and some Plugins will tidy-up and remove such trailing spaces. It will also look odd in Reports and Diagrams especially at the end of lines where the comma may wrap to the next line.

I hadn't considered reports, because I haven't got round to them yet.

Another way round the problem, which might please both sdes of the argument, is if, as one typed, a drop-down list appeared showing next three or four closest alphabetical matches from the list. This is a method used in FTM, if memory serves. I would only need to type as far as Step before Stepney appeared alongside Stepney Green. This would also aid distinguishing between such places as Kingston upon Thames and Kingston Upon Hull, both of which require a lot of typing before correct one is identified.