IRC Logs for #cmvt Monday, 2013-04-15

*** DanielBaird has joined #cmvt00:36
*** DanielBaird has joined #cmvt01:13
prologicHey Dan01:13
DanielBairdhi01:51
DanielBairdi finally have a leaflet map in my crappy layout, but i have to learn a bit more about angular "directives" before i can do anything about it01:52
prologicok no worries01:59
prologicI'm still working on clipping and masking of raster datasets02:00
prologicvia polygon(s)02:00
prologicI have a working implementation02:00
prologicJust cleaning up the code and creating a nice API02:00
prologicPerformance-wise:02:02
prologic0.6s for 5km02:03
prologic1.2s for 1km02:03
prologic11.5s for 250m02:03
DanielBairdso once i need to add layers to leaflet, do i jsut tell leaflet some vector layer url?02:30
DanielBairdor will i need to track the map zooming and panning myself, and refresh the vectors manually02:30
DanielBairdhmm actually it looks like vector "layers" in leaflewt aren't proper layers anyway02:32
DanielBairdleaflet has a geojson layer type but it takes a data object instead of a url.  so if we don't want to load the entire maps's worth of features in one go, i'll have to manually construct the url with bounding box info etc each time the map moves02:35
DanielBairdraster data is easier from my end, i just tell Leaflet the WMS url and it manages fetching the tiles as necessary02:37
prologicyeah03:18
prologicI can certainly provide you with a restful service to get at the vector layers03:18
prologicwe can make the API similar to WMS03:18
prologicyou pass me a bounding box03:18
prologicthe current map's extent03:18
prologicand I give you geojson that represents all visible vector layers that intersect that bounding box03:19
prologicI can do this quite quickly (I've found)03:19
prologiceven without rtree indexes03:19
prologicwhich I may add later on03:19
DanielBairdhmm i guess the minimum arrangement is that i just chuck away what i have and replace it with the vector features you send for the new bounding box.. ideally though i'd be able to keep ones that should still be onscreen, and just add the new stuff.03:41
DanielBairdi suppose we can give features a unique persistent id03:41
DanielBairdthat'd be enough for me to not re-add the ones i already have03:42
DanielBairdwhat vector features do we have, anyway?  for edgar we had crazy amounts of stuff like hundreds of thousands of occurrences for speceis03:45
DanielBairdi guess we have the region polys03:45
prologicyes03:59
prologicwe have IBRA region polys03:59
prologicand we will have NRM(s) and LGA(s) as well03:59
prologicnow I _have_ to simplify these03:59
prologicbecause some numty has gone around with a greak'n GPS03:59
prologicand walked some of these regions03:59
prologicsome are in the 400,000 points03:59
prologicand over 8MB in data to serialized to the client03:59
DanielBaird:) awesome04:00
DanielBairdif we were clever, we'd use simple versions until they zoomed in really close.. then in guess show the border segment they were looking at and fake the rest to be straight rectangular borders just off screen04:01
prologicyeah I know04:01
prologicsome people astound me :)04:01
prologicwhy you need an IBRA region to be that accurate04:01
prologicgod only know04:01
prologic+s04:01
prologicyes04:01
prologicideally (in the long run)04:01
prologicyou would pass me the extent you want04:01
prologicand I've dynamically work out how much to simplify the region by04:02
DanielBairdyeah noone who uses this tool will care about finer grain than 250m squares in any case04:02
prologicprecisely04:02
prologicsome of these regions contain more points04:02
prologicthan what you'd get at 250m04:02
prologicanyway that's worst case 400,000 points04:03
prologicI think LGA regions have similar issues04:03
prologicfrom the datasets I'm aware of in my previous job with old qovt04:03
prologiclargest region I think was over 10,000 points04:03
*** DanielBaird has joined #cmvt08:43
*** DanielBaird has quit IRC08:52
*** DanielBaird has joined #cmvt09:18
*** DanielBaird has quit IRC09:22
*** DanielBaird has joined #cmvt10:19
*** DanielBaird has quit IRC10:23
*** DanielBaird has joined #cmvt11:19
*** DanielBaird has quit IRC11:23
*** DanielBaird has joined #cmvt13:20
*** DanielBaird has quit IRC13:25
*** DanielBaird has joined #cmvt13:58
*** DanielBaird has quit IRC14:24
*** DanielBaird has joined #cmvt14:55
*** DanielBaird has quit IRC15:03
*** DanielBaird has joined #cmvt15:30
*** DanielBaird has quit IRC15:34
*** DanielBaird has joined #cmvt16:31
*** DanielBaird has quit IRC16:35
*** DanielBaird has joined #cmvt17:31
*** DanielBaird has quit IRC17:35
*** DanielBaird has joined #cmvt18:32
*** DanielBaird has quit IRC18:36
*** DanielBaird has joined #cmvt19:32
*** DanielBaird has quit IRC19:36
*** DanielBaird has joined #cmvt20:33
*** DanielBaird has quit IRC20:37
*** DanielBaird has joined #cmvt21:33
*** DanielBaird has quit IRC21:38
*** DanielBaird has joined #cmvt22:34
*** DanielBaird has quit IRC22:38
*** DanielBaird has joined #cmvt22:56

Generated by irclog2html.py 2.11.0 by Marius Gedminas - find it at mg.pov.lt!