Upgrade from AIO 64-bit 5.2.3 to current (6.0.6)
It’s about time I upgraded to the latest version (nothing in particular, just keeping up to date, or not getting too far behind!)
Windows 10 64-bit
I have made non-media backups
Question: should I let the first run of 6.0.6 do the conversion of the database(s), or should I create a new database in 6.0.6 and import the backup.
I have read Version compatibility[when upgrading from 5.1.3.2. to 6.0.6]? - #2 by DaveSch and comments by DaveSch which suggest the latter course of action. But I am just sampling the combined experience of the Discourse list.
I have 3 databases, a main one which contains the majority of my data, and two smaller ones which contain other families’ data which I don’t access very often.
Many thanks
GRAMPS: AIO64-5.2.3-r1-aa03f5a
Python: 3.11.9
BSDDB: 6.2.9 (6, 0, 30)
sqlite: 3.46.0 (2.6.0)
LANG: en_GB.UTF-8
OS: Windows
I have some reservations about the ‘purity’ of the Restore of a Gramps backup.
In the most obvious example, the standard Gramps IDs are recast using the current ID Formats in Preferences. (Which, in a new installation, revert back to the insufficient four-digit variants.)
If you have IDs mentioned in Notes, Descriptions, scripts or external web pages; then you probably want the IDs to stay exactly as they were saved in the XML export. But there is no way to make the Import that ‘literal’… it is a bit too intelligent sometimes.
The only IDs that re-import reliably are the fully customized IDs. (Such as where you used an external UUID instead of the ID formats that Gramps generates. I mostly have those in GeoNames validated Places and Person IDs that correspond to a particular Family Society.)
The other parts that are “at-risk” are those that reference addons and customizations by indexing rather than by name. (Added Rule packs, Filters, etc.) If the referenced item is not installed, then the reference might fail… and be purged by Gramps internal validation. So you want to install all the addons before importing (or converting) tree data.
Yes, these might be niggling details. But they also might only be the most visible differences. What else might we not discover until months or years later? When we have committed too many manual updates and additions to even consider reverting?