[Freeciv-Dev] [bug #18087] Increase max map size to 2, 048, 000 tiles (and max linear size to 32, 768)
Update of bug #18087 (project freeciv): Status:None = Fixed Open/Closed:Open = Closed ___ Follow-up Comment #23: Final commits for this patch ... ___ Reply to this item at: http://gna.org/bugs/?18087 ___ Nachricht geschickt von/durch Gna! http://gna.org/ ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
[Freeciv-Dev] [bug #18087] Increase max map size to 2, 048, 000 tiles (and max linear size to 32, 768)
Follow-up Comment #20, bug #18087 (project freeciv): I only noticed one [memory leak]: in sg_load_map_known()), SAVE_MAP_CHAR() - sg_failure_ret() - sg_check_ret() can return prematurely. I will try to find a way to handle such cases. Would it help to add a (possible) goto action to the macro, i.e goto CLEANUP? Having a standard goto CLEANUP would oblige every direct or indirect caller to have a CLEANUP label, and perhaps limit you to one use of this stuff per function, neither of which is likely to be acceptable. I'm not 100% sure what the answer is; probably something like adding a cleanup argument to the macro, which is executed as code before returning, and into which you'd put free(whatever). Anyway, I've raised bug #18347 about this. trunk-S2_3-hugemap-colatitude.diff It is only needed if the map size is increased further (patch 3). Thanks for confirming. Is there something like #error? It could stop compilation if the check fails ... #if/#error looks like the right thing (and would probably work in this case) but often it's subtly wrong or annoyingly incapable (no sizeof, etc). I've raised patch #2811 for a non-preprocessor static assert mechanism. ___ Reply to this item at: http://gna.org/bugs/?18087 ___ Message sent via/by Gna! http://gna.org/ ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
[Freeciv-Dev] [bug #18087] Increase max map size to 2, 048, 000 tiles (and max linear size to 32, 768)
Follow-up Comment #19, bug #18087 (project freeciv): I'm working my way through a lot of new messages. Furthermore all of my all patches have to be rebased to the current trunk - a lot happend in the last half year! trunk-S2_3-hugemap-stack-overflow.diff Thanks for working on this patch - this on looks good! I'm compiling current trunk at the moment. I only noticed one: in sg_load_map_known()), SAVE_MAP_CHAR() - sg_failure_ret() - sg_check_ret() can return prematurely. I will try to find a way to handle such cases. Would it help to add a (possible) goto action to the macro, i.e goto CLEANUP? trunk-S2_3-hugemap-colatitude.diff The changes in apgen.c only prevent division by 0 errors. If I remember correctly, the change in mapgen_topology.h tries to addaped the map generator for the bigger size (try a size of 32x500 or 500x32 ...). It is only needed if the map size is increased further (patch 3). trunk-S2_3-hugemap-biggermaps.diff Yeah, this is for trunk or for people who need such a big map (akfaew?). Aside: I think the correct way to express the relationship between MAP_MAX_SIZE and MAX_DBV_LENGTH is to publish the latter in bitvector.h and have a static (compile-time) assert in the map code. But we don't seem to have a static assert mechanism yet. I may look into it. Is there something like #error? It could stop compilation if the check fails ... ___ Reply to this item at: http://gna.org/bugs/?18087 ___ Nachricht geschickt von/durch Gna! http://gna.org/ ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev
[Freeciv-Dev] [bug #18087] Increase max map size to 2, 048, 000 tiles (and max linear size to 32, 768)
Update of bug #18087 (project freeciv): Assigned to: jtn = syntron Planned Release: 2.3.0,2.4.0 = 2.4.0 Summary: server segfault on a map with 1.3M tiles = Increase max map size to 2,048,000 tiles (and max linear size to 32,768) ___ Follow-up Comment #18: I've committed trunk-S2_3-hugemap-stack-overflow.diff to S2_3 and trunk, under the auspices of bug #17962. Autogame testing now in progress. (You might reasonably think it should be the other way round, but I want to maximise the chances of this patch being tested before release.) Anything that comes out of that, or any further comments on that specific patch, should be made on bug #17962, please. Since the segfault in the bug name has probably been fixed by this, I'll repurpose this bug for discussion of increasing the map size limits and anything that needs to be done to make that work, retarget it to 2.4.0, and reassign it back to Matthias. ___ Reply to this item at: http://gna.org/bugs/?18087 ___ Message sent via/by Gna! http://gna.org/ ___ Freeciv-dev mailing list Freeciv-dev@gna.org https://mail.gna.org/listinfo/freeciv-dev