Similar thoughts here, too. Thinking particularly of the Docker image, the main use case for them is ensuring a reproducible build on the user's side. Those using the `latest` tag (indicating they're not so bothered about versions) will start building with 3.6.1 automatically anyway. Those who have pinned builds to 3.6.0 presumably want to avoid unexpected changes - and given it's short life, I expect those are users who follow GDAL quite closely.
I also don't get the impression this is *significantly* worse than the normal bugfixes we see with any release - minor bugs like coordinate rounding and interpolation issues have affected plenty of raster datasets in the past, but we wouldn't consider a retraction of all GDAL versions up to that point. Regards, Daniel On Thu, 15 Dec 2022 at 20:11, Greg Troxel <[email protected]> wrote: > > Even Rouault <[email protected]> writes: > > > Question: now that 3.6.1 is out, what to do with the > > osgeo/gdal:xxxx-3.6.0 docker images: > > > > - keep them > > > > - remove them > > > > - alias them to the 3.6.1 ones which are now available ? > > Basically I'm with Bas on this. > > But even more strongly, do not alias 3.6.0 to 3.6.1. If someone asks > for 3.6.0, getting it is fine, and getting an error is ok, but getting > something else is not ok. > > As a packger I don't really care, because I was fortunate to still be > waiting to get around to 3.6.0 (part cmake, part life in general, part > being a little shy about .0). So I will, when I get to it, go right to > 3.6.1. > _______________________________________________ > gdal-dev mailing list > [email protected] > https://lists.osgeo.org/mailman/listinfo/gdal-dev >
_______________________________________________ gdal-dev mailing list [email protected] https://lists.osgeo.org/mailman/listinfo/gdal-dev
