-
Notifications
You must be signed in to change notification settings - Fork 64
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
ST_LineLocatePoint for geography and one point linestring crashes backend #354
Comments
I'm also thinking the fact that we have the same names of functions is going to pose a problem for people upgrading. Not sure the easiest way to fix that. It might be good to start thinking about using prefixes for your custom functions that don't overlap with PostGIS ones. For example pgrouting uses prefixes pgr_ and pgr so their functions never collide with PostGIS ones. h3-pg extension prefixes theirs with h3 and _h3 so their function names never collide with PostGIS ones. |
Regarding the naming of functions, all the functions added in PostGIS 3.4 coming from MobilityDB were designed for PostGIS and thus they are commented out with #ifdefs when compiled with PostGIS 3.4. |
Yes. The issue I see with that is, say a user is upgrading from PostGIS 3.3 to PostGIS 3.4 and they have MobilityDB 1.1 installed for PostGIS 3.3. If they try to do a PostGIS 3.4 upgrade before they do a MobilityDb upgrade, then the install will fail because of the new security features in PostgreSQL, will prevent replace of a function installed by another extension. I haven't tested myself, but I suspect that is what will happen. So now you have a dependency order of upgrade which is pretty annoying. |
We finally decided to remove the functions from MobilityDB. It would simplify everything for both PostGIS and MobilityDB. Users that need these functions would be required to upgrade to PostGIS 3.4 but as consequence will have available all goodies in the latest release. |
As an aside, we did a revision of the geography functions now in PostGIS 3.4.
|
Thanks. I'll ticket on our end for a fix before our 3.4.0 release |
When packaging PostGIS 3.4.0 for Windows, please use our alpha release |
I've ticketed on our end - https://trac.osgeo.org/postgis/ticket/5460 Regarding the v1.1.0. I think I'm having some issues with it, but will get back to you on that once I figure out if it's something wrong on my end. Last I tried last week, it compiled, but was giving a "NOT a WIN32% application" or some such thing when I went to load (even with making sure PostGIS was in loaded libraries). I haven't seen that error in a while. This was against PG16beta / PostGIS 3.4.0 and latest mobility develop branch. I'll put in a separate ticket if I need help on it. |
Dear Regina |
I am sorry for that. I didn't know. I will do as instructed next time. |
I think pramsey caught it as all those changes already seemed to be in place. I think you can close this ticket out unless you are waiting for your release before you close. |
Many thanks Regina !!! |
Describe the bug
PostGIS garden test triggered a crash on ptarray_locate_point, which is I think mostly a copy of the MobilityDb version.
Details here:
https://trac.osgeo.org/postgis/ticket/5456
backtrace shows crash around here:
https://github.com/MobilityDB/MobilityDB/blob/develop/mobilitydb/src/point/geography_functions.c#L595
I can trigger the same error with MobilityDb's function on my PostGIS 3.3.3 instance.
To Reproduce
Steps to reproduce the behavior:
Expected behavior
I'm not sure why this should even be allowed? Probably should just error out if you pass in a linestring that has only one point.
Specifications (please complete the following information):
Use the commands:
output
The text was updated successfully, but these errors were encountered: