10 replies to this topic

[Alfred]

You can type in any number of generations and get the correct report except for the numbers displayed.

Because of the way binary numbers and computers work, the numbers start to get messed up after 31 generations.

In a future program, the field could be widened to hold twice as many generations or even more, but right now, the number is a double word, 32 bits wide, one bit is the sign, negative or positive, so the ahnentafel numbers get turned nagative at the 32nd generation and start overrunning past that.

So, the wish has to be for a wider field if we want these numbers to be correct past 31 generations.

[apw6270]

That was right - you only need 5 bits to store 32 numbers, not 32 bits, therefore the limit is artificially imposed.

[kbens0n]

Must be more to it... than just being a computer mathematician!
If it was as easy to implement, as you assume... it would have been done a long time ago.

[RootsMagician]

That was right - you only need 5 bits to store 32 numbers, not 32 bits, therefore the limit is artificially imposed.

But you need 32 bits to store the ahnentafel (ancestor) numbers of the people in the 32nd generation, therefore it *isn't* artificially imposed.

[apw6270]

I don't assume anything and it is simple to implement, and before you say "go on then, do it", I already am.

There is no reason to store the ancestor number in a single bit, therefore the limit *IS* artificially imposed.

It's all down to poor or lazy programming.

[Alfred]

first generation, 1 person
second generation, 2 people
third generation, 4 people
fourth generation, 8 people
fifth generation, 16 people
sixth generation, 32 people
- - - - - - - - - - -
Take that on for 32 generations and you have a total of 4294967295 people who need to have unique ahnentafel numbers.

Can you explain to me how you can keep track of all these with only five binary bits.
If you can do this, you had better get a programing job, because no one else can.

You seem to be talking about the number of generations and the rest of us are talking about the number of people.

[John_of_Ross_County]

But you need 32 bits to store the ahnentafel (ancestor) numbers of the people in the 32nd generation, therefore it *isn't* artificially imposed.

I once worked for company that was acquired by another contractor. A personnel related program was written in DEC PDP11 assembly language with some reports in Fortran. Employee badge numbers were expressed as single precision signed 16 bit integers. Historically, badge numbers had ranged from 1 to about 10000. The new employer added 50000 to everyone's badge number to match their other locations. The magic number for the largest positive single precision integer is 32767. The assembly language programs were OK, but the Fortran programs blew up since this version of Fortran interpreted everyone's badge number as a negative quantity.

Now more to the point, how many data structures would be impacted by RootsMagic switching to double precision integers? But if you made this switch, then someone would want more than 64 generations.

[apw6270]

Clearly it is not possible to use just 5 bits to store 4KM numbers, and I haven’t suggested that it is.

Clearly the number of ancestors in a number of generations is 2ⁿ where n is the number of generations.

If one ancestor was added to the database every second, it would take over 136 years NON STOP to add all 4KM of them. No one could possibly add all the ancestors in 32 generations.

The limit of 32 generations is imposed because someone who has only half thought about the problem has decided that 4KM ancestors would be reached before 32 generations.

I guarantee the originator of this thread has *significantly* less than 4KM ancestors in his 32+ generations.

The limit of 32 generations is artificially imposed.

[RootsMagician]

The limit of 32 generations is imposed because someone who has only half thought about the problem has decided that 4KM ancestors would be reached before 32 generations.

The limit of 32 generations is artificially imposed.

That might be true *IF* a numbering system was used that only allocated a number to people who actually appear in the report. But the standard ancestor numbering system used in genealogy is the Ahnentafel numbering system which assigns a number to every possible ancestor so that doubling a person's number gives their father's number (and adding one to that gives the mother).

RootsMagic uses the standard Ahnentafel numbering system, and thus is required to use the full 32 bits to handle numbers in the 32nd generation.