A quick logon is need for Ancestry that simply logs on and which does not do any sort collecting of changes or comparing of results. The reason is that the standard logon can take a very long time and 99.99% of the time the only reason I need the logon is to be able to see Ancestry WebHints.
Good point. the logon to family search goes to a screen of choices rather than a forced listing of people and changes for the ancestry login. It should also happen automatically like it used to.
Automatic would be nice, too. For reasons unknown to me, a decision appears to have been made to withdraw support for the automatic logon. But for me as a sample size of one, a fast logon is much more important than an automatic logon. I would be happy to do the manual logon if it would just logon without collecting changes at the same time.
Indeed, I’ve asked for this as soon as they changed it. Now that my laptop will not finish data collection at all it’s even more important.
I’ve seen data collection stop because of to many unmatched people. It will happen if you start adding a lot of people to your database and don’t upload them in TreeShare. The only option then is to upload/download a new tree.
I never match people anyway. Instead, I always upload a new tree on a periodic basis.
People who are doing the matching will surely do the collection on a regular basis no matter how the open works. They would just do the open, and then at some future point they would click on the Ancestry icon to do the collection. The collection could take place a few seconds after the open or a few minutes after the open or a few hours after the open. It wouldn’t matter.
There are other reasonable options that could be implemented. One possibility would be to support an Open With Collection option called Ancestry TreeShare and also an Open Without Collection option called Open Ancestry TreeShare. Another option might be to restore the automatic open, just making sure that the automatic open does an Open Without Collection. But to me the simplest and most logical option is to deconflate the Open and the Collection.functions. Publish => Ancestry TreeShare would do the Open function and the Ancestry icon in the main left tab bar would do the Collection function. Or really, I don’t care as long as I can open without Collection.
I have a ticket in at the moment where Data Collection is stopping (#129070) and takes several minutes to get to that point. Has nothing to do with number of people unmatched, although it may have to do with # matched as I have over 20K matched. In any case, it’s annoying to have to go through that process just to see hints.
If you upload a new tree to ancestry you will get all hints from ancestry again, and you can not distinguish between hints you accepted already and really new hints from ancestry that you have not seen yet. But I think you work with the hints from ancestry.
How do you solve this?
I have 60k matched and the process is always remarkably fast and efficient - unlike the process for the first download which is very slow and stops apparently at random.
Yes, there’s something fundamentally broken in the process. On my Mac Pro (2013) it takes 12 seconds to do all the data collection. On my older MacBook Pro (2019) the same file would never complete Data Collection, even if left overnight, my new MacBook Pro (2023-M2) same thing. I even took a freshly imaged MacBook Air (M1) from work and installed only RM9 on it and saw the same issue. All timeout after 3-4 minutes. Now you can cancel out of the data collection and still get hints, but it’s an annoying and unnecessary step. Just auto-login like with Family Search. I know it’s an issue with Ancestry like RM 7.7 when you’re somewhere without internet and it asks every 15 seconds “cannot connect to Ancestry” but even that it pointing out an issue with how the API is used.
Basically, I don’t solve this. That’s not really the way I use RM’s WebHints.
When TreeShare was first introduced, I quickly realized that based on the way I work that I would never be able to keep my RM database and my ancestry tree in sync. I had an RM database of over 60,000 individuals at the time and I was spending more time trying to do the manual sync than I was doing actual research.
Also, I found lots of problems where I could never make my data in RM match my data in ancestry because the data models between the two systems are very different. Indeed, when I looked carefully at my RM database and my ancestry tree immediately after my initial upload from RM to ancestry, I found humongous numbers of people who had mismatching data. TreeShare removes them from the changed list, but that doesn’t mean that they really match. There is not a bug in TreeShare. People in RM don’t match with people in ancestry because they can’t be made to match.
In addition, I had this idea of chasing down every hint there was. My idea was that I would either reject each hint because it was wrong or else accept each hint and incorporate the newly found data into my RM database. But I couldn’t figure out a way to manage the actual hints because I had tens of thousands of them to chase down, maybe even hundreds of thousands. There is no way to search for people with hints or make groups out of them or anything like that because the hints are not stored in the RM database.
Remember that RM supports hints for ancestry, familysearch, myheritage, and findmypast. I chose to use hints for ancestry, familysearch, and myheritage. I was already paying for a subscription to ancestry and I already had a subscription to familysearch which is free. I found that the ancestry hints and the familysearch hints overlap a great deal but that each had data the other didn’t have. I tested the hints with myheritage and I found enough value in them including data that was not in ancestry or familysearch that I got a paid myheritage subscription. I tested the hints with findmypast and I didn’t find enough value in them to get a paid subscription. There is nothing wrong with findmypast. I think it’s just that since it was the last one I tested, there was little or no additional value to be gained from yet another paid subscription.
The point being is that I couldn’t base my strategy of managing RM’s WebHints just on TreeShare and ancestry. So I needed an overarching WebHints strategy. I therefore decided not to try to chase down every WebHint in the world nor to stress over whether hints were approved or not or anything like that. I type everything into my RM database by hand anyway, rather than using TreeShare or RM’s direct interface into familysearch to move data into RM. So I just look at the WebHints for the people that are currently on my RM screen and use them as additional information for my research. I sometimes reject hints but I seldom even bother to approve hints anymore. I truly just don’t care about the approval process for WebHints. I just want the data.
To tell you the truth, I seldom even click on the WebHints icons anymore to get ancestry hints and familysearch hints. Ever since RM9 introduced the new feature to right click and Jump To a person in ancestry or familysearch, I usually just right click and Jump To in order to see the hints. I make heavy use of RM’s Descendant View and I was bummed out that RM9’s Descendant View doesn’t show the WebHints. But the right click and Jump To works just fine in Descendant View. Of course, the Jump To doesn’t work for MyHeritage so I still need to click on the WebHints icons for myheritage hints.
I realize that the way I work may be different than the way many other RM users work. But that’s one of the many strengths of RM, that it’s so flexible that there are lots of different ways to use it that are all good.
@thejerrybryan I completely understand the logic of your request for a log on to Ancestry that doesn’t download treeshare changes, but I wonder how big a problem this is for you.
I log on a lot and I have recently taken to timing it; the process of downloading all the changes from Ancestry after entering my username and password takes about 10 seconds and doesn’t vary very much. There is also no significant difference between the time to log on and to start treeshare when already logged on. This is on a modern windows laptop using the 64bit windows version of RM9 a good wifi internet connection and with a database of 60k people. I usually have 100-200 changed or unmatched people each time I run the process.
Are you asking for a change to save you 10 seconds each time, or is your experience significantly different? @keithcstone also wrote about the process taking some time and @rzamor1 noted that a large number of unmatched people might cause problems. It might be interesting to compare notes.
The logon takes about 1:55 on RM9. It takes about 5:45 on RM7. That’s a lot longer than 0:10. It’s interesting that RM9 is actually a good bit faster than RM7 for this particular function on my particular RM database and on my particular computer. I have about 40,000 people in my RM database these days, down from about 60,000 a couple of years ago.
It’s the same Ancestry tree for both RM7 and RM9, and I keep my RM7 and RM9 database in lock step with each other. I’m not sure how many people are on the TreeShare changed list, but it looks like it’s just a few hundred. I delete my Ancestry tree and upload it again it from RM on a regular basis, so the changed list never gets especially long, even though I never actually do the manual sync that is supported by TreeShare.
My machine is a fairly old Windows laptop with four i5 processor cores. But I have long since upgraded the memory to 32GB memory and the disk to a fast SSD disk. I also have fast Internet… During the TreeShare logon, RM is CPU bound, maxing out one of the four processor cores. During this time, disk utilization is not a bottleneck, memory utilization is not a bottleneck, and network access is not a bottleneck. Indeed, the disk and network utilization is almost zero. The bottleneck is all in the CPU processing.
I hadn’t used RM7 for some months so a) I had forgotten how long this process took and b) I now have a large number of differences between my RM7 database and my Ancestry tree.
Using RM7, the process did take some minutes (I got interrupted by a phone call and lost count); it was also divided into two stages, 1) collecting data and 2) comparing/looking for duplicates. Both of these two stages took quite a long time and, as you say, mainly used CPU.
By contrast on RM9 (with two databases fairly closely in line) I never see the second of these two stages and the first completes in c10 seconds.
You don’t mention how much of the time in your two counts was spent on stage 1 and how much on stage 2. I would find it quite hard to explain why the 1st stage would take much longer on your RM9 database than mine; although my PC is fairly modern, I did not splash out on the fastest cpu. I have an AMD Ryzen 5625U processor with six cores, only one of which is being used by RM9. I have also chosen the option to prioritise minimising power use even at the expense of processing speed.
I have been struck for some time by the speed of the RM9 process. I wonder how it works. If I choose to, I can use Treeshare to see all the facts I have for all the people in my Ancestry datbase with an outline of the sources for each fact. Do they all get extracted and downloaded each time or does RM store the Ancestry information and only update the changes? Either way, it is quite impressive that Ancestry can extract and send the data so quickly. I assume that RM stores the data in a format similar to that sent by Ancestry, so that there should not be much processing involved in storing what Ancestry sends it.
From what I can see the number of unmatched people has nothing to do with the time (or very little). Neither does network. This maybe a Mac specific problem as I haven’t had time to test in on a PC. I can take the exact same file and on my main machine (Mac Pro 2013, 64GB) and it’s roughly 12 seconds, sometime a minute when Ancestry is feeling grumpy. On my previous MacBook Pro (2019, 32GB) it would never complete, even if everyone had been previously matched on the other machine. As a test I even took a spare MacBook Air M1 from work and imaged it with my credentials and installed only RootsMagic on it and the same thing. This is either wired or wireless. On top of that, I often need to quit RM to work in SQLite, such as clearing deleted Family Search Ids, so then I need to log in again. That means that I may have to go through this process 4-6 times in a given session.
I’ve had time to do more research. I’ve copied the file and media folder to a fairly beefy Windows 11 PC and after an initial login of roughly 5 minutes it completed data collection. I did notice after the first data collection took nearly 5 minutes subsequent attempts to log into Ancestry took roughly the same time as on the MacPro. This is with a file that was in sync prior to copy.
I suggest the issue lies in some file or folder that’s initially created during initial TreeShare operations. This file / folder was always there on the MacPro 2013 as it was the one I always used RM 8/9 on. The older MacBook Pro which worked with TreeShare on RM 7.7 also had the file / folder. When I removed RM 7.7 and RM8from the MacBook Pro, upgraded the OS, and installed 9 that file / folder was removed was isn’t able to be recreated in a timely manner. That is the case on the new MacBook Pro 2023 and the older MacBook Air M1, it’s also the reason it took nearly 5 minutes for the first data collection to be completed on the GeekOM Windows PC.
I still want auto-login like Family Search though.
Renee, I’ve done even more research. During the data collection process the file:
is created and removed as soon as data collection is complete. On both my Mac and PC where TreeShare is working the file AT_63317064.DT8 in the same folder is about 226MB. On the Mac where it is not it is 53Kb.
If I copy the AT_63317064.DT8 file from the Mac that is working to the Mac where it is not the Treeshare data collection completes in a few seconds, so obviously the issue lies with the AT_63317064.DT8 file. So people experiencing long delays in data collection are having an issue with management of that specific file.
Ancesty Member Tree (AMT)
SQLite Rollback Journal Temporary Files Used By SQLite for database transactions, not Ancestry login and possibly not for initial API data exchange.
Like you, I see the two stages in RM7 and I don’t see the two stages in RM9.
I just reran my RM7 logon to Ancestry with the result that stage 1 was 0:51 and stage 2 was 4:41.