TMG-L Archives

Archiver > TMG > 2002-01 > 1009908408


From: Richard Brogger <>
Subject: [TMG] Merging and splitting.
Date: Tue, 01 Jan 2002 12:06:48 -0600
References: <14d.69318f7.29632763@aol.com>


wrote:
>
> is easier and less complicated in TMG to separate out people
> from a data base than it is to merge smaller ones into larger ones. Or so the
> experts, which I am not, seem to say. Good luck-Dale

Hi Dave,

You dodged that very well. <g> The getting started manual says it is
easier to split than to merge. It may be true for some data sets but
it is totally false for others. In my own data set I split a 2g
grandmother's branch from the Brogger line. There were four places
that the two lines were tied. It was a simple matter to break those
four ties, set a flag and make two new data sets.

I inherited a 120,000 person data set. I would like to split off all
branches with >1000 people. How does one do that in a data set that is
not well known? I have tried to accent the main trunk and run ancestor
reports on the spouse. Sometimes I get a list of just a few persons.
Sometimes I get a large list but on the following passes, I get the
rest of the data set. i.e. there is an additional tie(s). I track down
the ties and break them. Run the report again and find that I only
have 70 people in that branch. Too few to make a new data set.

In the time it takes to track down the fact that the branch is small,
I could have merged ten data sets into one.

I have made trial splits between a couple and or their children. Made
a GEDCOM. Imported the GEDCOM into GenViewer and looked for "islands".
I have imported a GEDCOM into FTM and tried to used the All-in-One
tree to make a split. Making a GEDCOM in TMG is slow, making a
All-in-One tree in FTM is so slow it will not finish while I sleep.

I have managed to split enough times that the main data set is down to
about 60,000 but my experience with merging and splitting does not
show that splitting is easier. For that statement to be true, one must
know where to make the splits. If TMG has the means to locate a
logical place to make a split, I have not found that means.

Users want TMG to have every feature we can think of. We want it to
handle huge data sets. We want it to carry the extra baggage
associated with 32 bit programs. We want it to be as fast as
GenViewer. We want TMG/FTST to run fast on the obsolete computer we
have. We are asking for something that is practically impossible.

I think that the features in FTST are great and I want more but I know
I can not expect a host of features and speed and still handle huge
data sets and do it on my present computer. Something has to yield. I
could buy a computer with multiprocessors running at today's best
speed, with a ton of memory but I find small data sets more practical.

As a FTM user, I kept splitting my data sets for several reasons. I
learned to handle most of the minor problems of having multiple data
sets. They are not insurmountable, especially with the features of
TMG4. FTST's ability to allow multiple data sets in one project
removes more of the problems. I have a few 32 bit genealogy programs
on my computer. ALL of them are slow when the data set is large. NONE
of them have as many features as TMG4 has, much less what TMG5 will
have.

Eventually, I will get my monster data set down to the point that I
can get acceptable performance. Some will buy faster computers to get
acceptable performance. Some will accept the slower speed. Some will
stay with TMG4. Some will expect TMG5 to do what is practically
impossible and then complain when it doesn't.

Richard Brogger

This thread: