I’m Christian, the maker of OpenAndroMaps.org.
I came across a fouty monster Relation with 29000 Members = Tongass National Park
Edit ist close to impossible couse evry tool I know quits do to the size of the relation.
redering for maps takes ages.
This relation is tagged as type=multipolygon (its a boundary in fact)
This enormous amounts of members is caused by the fact that all single islands are menbers = the national park ends at every shoreline = IMO thats silly.
For me this relation have to be deleted and recreated wit the outer borders only.
However, thats your OSM region, so…
Not sure what exactly you want to edit. If you just want to change the tags of the relation there is no memory problem (at least not in JOSM), just don’t download all members.
If you want to edit the full relation though you probably need a 64bit JRE and a JRE option to increase the available memory. Try this on the command line:
I’ve been “doing” these for at least ten years, and a few years ago I got a new machine: a 4.2 GHz (bursts to 4.5 when it has to) Quad-Core Intel Core i7 with 64 GB 2400 MHz DDR4 RAM. I use Java, which is “a correct” solution (there may be others, but its tweak-ability on the command-line is exactly how I launch JOSM with “fatter” heap-space memory footprints). From my command line, I do so with:
(meaning 2048 megabytes or 2 GB of heap space for Java, and with “/Applications” you might know this means macOS, version 11.2.2 to be exact).
It took over two full minutes (via a 100 Mbps 'net) for JOSM to download this relation, but it did. And my fan ran and my CPU pegged and then everything calmed down as JOSM displayed the relation (rather prettily, I must say). That does seem to be a record for lengthiest-single-item-download-time in my almost 12 years of OSM. I didn’t have to bump up the RAM beyond 2 GB. JOSM reported the relation has 29194 Members.
I am curious what technology stack you’re using that has such difficulty with the Tongass relation. It seems like it renders just fine in OpenMapTiles, as evidenced in this openstreetmap-americana rendering: