retsof

2GB of RAM seems awfully tight. I don't know whether Win 8 needs more memory or not but more always helps and memory is fairly cheap these days. What you would have to get depends on what kind of slots are available in there. If there are 2 slots and two 1Gb sticks then they should be replaced. If 4 slots with one 2Gb stick or two 1Gb sticks used, then one 2Gb stick or two 1Gb sticks could be added.

4 Gb is the max for a 32 bit OS. Some could be taken for graphics, etc. so you don't get all of that, either.

Cautions are advised to buy memory with the same type, speed, timings and latency as existing memory (questions for the computer store).

It could be swapping things to disk all over the place, leading to those "stopped responding" errors. You could not isolate install writes from swap writes by just looking at the disk light.

Another program could also be chewing up cycles, leaving little for TMG. The task manager could help find problems of that type. Don't run the install at the same time as a full system virus scan, in other words.

I have 4GB on a desktop running Vista and 16 GB on a desktop running 64 bit Win 7. (TMG remains as a 32 bit application in only one core.) TMG7 runs fine with Vista, but I have SQZ restore issues in another thread with TMG8, so I am not going to put it on a portable laptop for now.

This is some good news I was wondering about and now I do not have to ask.

Since V8 is causing problems for me (in another V8 thread here), I want to put V7 on the faster computer I was originally intending to use for V8.

I have the registration codes for BOTH V7 and V8.

I am getting closer.

I backed up the V7 file and did a restore to V8.

The primary birthdates all show up in the picklist now.

An earlier V8.00 restore could have had problems but showed no outward indication by message.

On the other hand, V8.04 now says that the memo file was missing or invalid and there are some other problems like a STRINGSFLAG alias that cannot be found.

I backed up the V7 file and did another restore to V8.

The memo file is still missing or invalid.

I am now running the optimize, VFI and optimize on the original V7 file and will back it up again. (I really had not made that many changes since I started trying to migrate to V8.) No problems were found by the VFI.

Both V7 and V8 are on hold for data entry and cleanup until something seems more stable, and have been so for about 6 months due to this problem.

The backups have been made with normal compression.

SAME ERROR with missing memo file.

I will backup V7 again with compression=none. The SQZ algorithm has caused problems for me in the past, especially during a numbered release conversion from V6 to V7.

SAME ERROR so I am kind of stuck now. The restore seems to run a little differently in each minor version.

If I ignore the error I can see the records with all of my primary birth dates in the picklist. HOWEVER there are no memo records (I compared one from V7) so I may make some updates in V7 and forget about V8 for now.

Why were all of the primary birthdates in the picklist in version 7? The identical record sets had primary birthdates and primary deathdates for each.

That could be the problem, and adding a primary tag to a birthdate does fix the picklist and show it for that record. (Good catch!) They USED to be there in V7. WHY the primary tags were removed I don't know. It appears that the version 8 conversion left one primary birth date there for a set of identical records, and removed it for second and subsequent identical records. Was that really a good idea?

Yes, I had run optimize, VFI, and optimize several times in V8 when I first noticed the problem.

If I can find two identical records, I won't bother to add a primary tag to the second. I will simply merge them.

I have seen this error in all versions so far of 8: V8.00, V8.01, V8.02 and V8.04.

SOME birth dates and death dates appear in the picklist and some do not, even for the same individual with duplicate records. I have many duplicate records I am trying to merge from different GEDCOM imports but this makes it a lot more difficult. I can match several on surname. Some show the birth dates and death dates and some do not even though one or both birth date and death date and also sort date are actually in the record. Updating the record and changing the birth date does not make the birth date appear in the picklist, so I do not know what is causing the problem.

When birth date or death date is used as the primary field they are sorted, but there could still be a lot of the blanks in the pick list. It is more difficult to analyze when I am only looking at thousands of blank birth dates in the picklist. I did check one sorted by primary birthdate that had the same surname, given name and a blank birth date, and the birth date is in the field.

Everything worked fine in V7+. Something in the search picklist changed in V8.

No, the width of columns on the search page is the same.

I also went into the search and dragged the columns wider. The ones that already had a date were there, but others with dates in the record did not show them in the search.

I later also dragged the widths smaller. Even at a width of only one number wide, I could still see part of a birthdate or deathdate, if the record had one.

After the merge of records I inspected individually, the birth date and death date were added.

After an optimize, VFI and optimize, the birth date and death date were removed.

It is all somewhere weird. At least data isn't being changed in the record itself, just the search.

I can get the dates added to the search if two identical records are merged and the record ID changes. Playing with the date field or some other data does not make a difference. Birth and death dates stay blank in the search, which sort of makes the search useless if checking for birth dates.

I am searching the picklist by surname, given name and birthdate.

The search display has surname, given name, birthdate and deathdate displayed.

After the optimize, vfi and optimize with no problems found,

a new search has the birthdate and deathdate REMOVED from the search, but is still in the particular record I had merged.

"Search" is an icon with a picture of a binocular.

Mouse rollover is "Picklist of people".

There are also a couple of choices for picklists, and this is the simple picklist.

The search worked correctly for this and other projects in V7, but this issue appeared with V8.00, V8.01 and V8.02.

I updated 8.01 to 8.02 successfully but an issue is still there.

In the search, some records indicate birth and death dates and some do not.

In a small project with identical records, both contain the same birth and death dates.

The seach by name and birth date record had one with name only, and another record with name and birth and death.

I merged these two records without deleting anything.

After that the search had the two duplicate entries of name, birth date and death date, as it should.

I can now delete the duplicate entry, but the search is lacking.

In the meantime, I will run an optimize, vfi and optimize to see whether it makes any difference. It didn't before, and I have only updated a couple of records.

I am not messing with my primary project at all now.

Reports: The change log says that some reports are still hidden while some issues are being worked on.

I updated 8.01 to 8.02 successfully but one of my own issues is still there. It is referenced in another thread so I will go there.

Something is not right at all.

Birth dates and death dates do not show up in the search.

A few of them are there for some reason, but not all.

However, if two identical records are merged, they are in the search.

This was even after the optimize, VFI and optimize.

I am reluctant to carry on with any activity until V8.01 gets more settled.

I got an email like this which contained the new serial number for V8.

support@whollygenes.com Serial Number for The Master Genealogist V8, Gold edition.

Windows 7 compatability is listed as a choice if you pick the install option to install a previous version (from a Windows backup).

When picking a version from the troubleshoot, Windows 7 compatability is not listed.

It was getting late last night, but I think I saw when picking Vista Service Pack 2 compatability that it insisted on going ahead with Windows XP compatability.

Something is still ugly somewhere in V8.01. I have a project restore issue in another thread, but I won't repeat it here.

It has probably been a 32 bit version all along, even on a 64 bit OS, so ignore that thought.

I thought of another way to check the integrity of the backup.

I restored the exact same V7 backup above to V7.04, which is still available, and it restored perfectly.

This morning, I found the option that said to install previous versions. There weren't any previous versions available, but there was a Windows 7 compatability listed. When trying to troubleshoot above, there was no Windows 7 compatability available.

So I notice.

Perhaps 8.02 will be better.

It installed in the (x86) path now. (for 32 bit version)

In other words, how do I get Windows 7 back to the 64 bit version.

I ran the tmg8setup again, but it still kept the 32 bit version, and says 8.01.

One project worked correctly, and was a very large one.

A medium size project with about 102,000 records has had some issues.

The memo file was invalid or missing, and I ignored it and went on.

When first displayed, it only shows name information. It knows about dates, as they are in the search, but not in the display.

I ran an optimize, VFI and optimize, and it found about 3 problems per record and dates were displayed.

The same problem happens in both 8.0 and 8.01, in both 32 bit Vista and 64 bit Windows 7. (I have 2 machines to test this on, but would prefer to use the WIndows 7 machine going forward.) I am trying to get something working on either. I have other issues in a different thread in Windows 7.

I tried two different backup files from different days, with the same results ... missing or invalid memo file.

I have had problems with the SQZ compression in the past when converting V6 to V7 and have lost data.

I have been using average size compression, but should probably go to compression = none for integrity, even though the backup files would be a lot larger in size.

I also tried to update 8.0 to 8.01

It has sort of resolved itself, but mutated from the 64 bit version I was using in Windows 7 to a 32 bit version.

While trying to troubleshoot the version with OCX not found as above, Windows 7 is not listed as one of the versions for compatability. Vista Service Pack 2 is the newest.

I tend to adjust my the ID numbers of my own close direct line ancestors to match the reference numbers ... UP TO THE SIZE OF THE DATABASE (ID number given for an add default) because I refer to them so much.

Others, both direct and indirect can fall where they may. The TMG Ahnentafel report always makes a handy reference.

BTW- usage of bizarre work-arounds, like nines complement years, fall apart for those born BC, then had events AD, like death, etc. Happens often.

 

Best wishes,

Here's Antonia the Younger of the Roman Empire. 9s complement of the sort date sets the timeline in the correct order in TMG 7 for me.

sort | date

9963 BC circa 0036 BC birth of Antonia

9984 BC 0015 BC son Germanicus Julius Caesar born

9986 BC 0013 BC dau Julia Livilla 'the elder' born

9989 BC 0010 BC son Claudius I Emperor of Roman Empire born

circa 0037 circa 0037 (AD) death of Antonia

adding "BC" in the sort date seems to be optional. It will work either way for some distance BC.

There must be some year in the future that is the high end for TMG.

After a quick test, I'll call it 3000 A.D.

3001 is recorded as an irregular date.

Here's something else though. Birth and death dates seem to be moving around to their correct spot but dates as recorded in notes do not. Something is different.

I may repeat a few trials and report later. The database seems to have a mind of its own today and even the above example is being resistant to change. I'll optimize / validate file integrity / optimize and do it again. I had imported a few records and had optimized but not a VFI today.

That might have fixed it. A note for 100 BC and complement 9899 BC moves to before the others.

p.s.

Ashleigh Brilliant did wonder whether the folks in the Dark Ages had problems dealing with the Y1K crisis, when AD years went from 2 to 3 digits.

http://www.ashleighbrilliant.com/writings.html#Y1K%20CrisisS

I have a few nines complement dates in an imported database, but they combined it with the regular date and didn't suffix it with BC to force an irregular date (since they probably weren't using TMG). I'll try it with that addition by nines complementing the sort date as above. That should be much better and I can put circa, before or after in the date fields and the sort will still be correct. It's easy to hit the "update sort field" by mistake so I also am putting the 9s complement in the memo field. It's a bit of double work, but might serve to show what I have entered into the sort field without looking again each time. Yes, the events for a person are visible in nice descending BC date order and cross over correctly to ascending AD.

Never mind about the before 0001 AD category. I lost the BC association in the picklist since the memo column isn't visible.

TMG does not handle years before 1 A.D. and struggles with the first century A.D. years before 101, as it wants me to "fix" the century, but finally accepts a year from 0001 through 0100.

There was no year zero, which messes up the continuity, but we are aware of that. 46 B.C. (year of confusion of 445 days before the Julian calendar started in 45 B.C.) introduces more fun, along with inconsistencies of setting early Julian leap years.

I put B.C. dates in the memo field as do most people. I did find that I can separate these from other unknown dates or no date at all by entering "b 0001" (before 0001) in the date field. That helps a little and gives these a catagory separate from later A.D. years at least.

I don't see any reference to what version 8 will do about this, but it's still early. The preliminary file of what it will do is still in the overview stage.

Gender or sex gets more confusing as it goes. Consider hermaphroditic beings, which have elements of both.

The man who photographed our wedding went to Canada a few years and got reassigned (sex change operation) as a woman. She and her wife are still living together.

TMG will not deal with a marriage unless between a man and a woman. Some states allow lesbian or gay unions. I have one friend in a polyamorous marriage (containing 2 men and 2 women).

There are polygamous religious colonies, and the reverse is possible in some societies.

Have you changed the compression parameters from normal to none or most to get anything different? It may not matter much unless your database is large, as this one is. A validation itself runs about 6-7 hours on one core of an AMD 3.2GHz Athlon X2 computer. Optimize is 20-30 minutes each. I am merging many duplicate records from some assorted GEDCOM imports, but it is very time consuming.

It seems counterintuitive to not run a backup before a long maintenance run, but it seems that is what I must do. Backups only work on clean databases. That does seem to make sense, since Dynazip seems delicate.

Optimize/Validation/Optimize is in progress from a recent restore of the recently optimized database. Validation fixed a few errors and is clean. It seems to be running better today. Yes, the validation kicks up the size of the database, and optimization pulls it down again to where it was the first time.

If the validation has a problem, I had also not thought of running an optimize before running another validation.

I found a new problem that seems to interfere. The notification screen for the automatic Microsoft update usually seems to be around when Foxpro stops working. (Vista Ultimate) I changed the setting to let me decide when to download something from Microsoft. Perhaps I get get through a good run again.