Migrating TIGER from Oracle to POSTGIS

['Sandro Santilli']

On Thu, Mar 20, 2025 at 04:42:54PM +0000, Ayorinde O Adesugba (CENSUS/DITD FED) wrote:

> Do you want me to submit a PR, so we can start reviewing and providing feedback?

Yes PR is a good idea. I think I saw a reference to it somewhere,
maybe add a link in this thread for easy reference in the future.

> I also need to discuss how the upgrade path will work out.

I guess the biggest decision to take is whether to support multiple
"formats" of topology schemas or "adapting" any pre-existing topology
to the new format.

At the moment we do have some support for multiple "versions",
in particular we allow for topologies to have a Z value or not.

Supporting multiple datatypes for identifiers would requiring having
to check which version a topology has in before reading and writing
those numbers, which may be not too expensive for the C functions but
worth profiling.

Pro of supporting multiple identifiers size would be to keep topology
small when you don't need large identifiers.

Con is having to maintain multiple code branches.

> I think
> the FACE, EDGE_DATA, NODE and RELATION tables are the 4 that need to be
> updated to allow INT8 IDs, but I maybe overlooking other upgrade items.

Hopefully our testsuite will help with finding this out, although I'm
afraid we don't have tests that use topologies created BEFORE upgrade
- but the testsuite does support pre-upgrade and post-upgrade hooks
being arbitrary SQL files so we could add what's missing.

--strk;
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 659 bytes
Desc: not available
URL: <http://lists.osgeo.org/pipermail/postgis-devel/attachments/20250326/30f1d270/attachment.sig>