Where communities thrive


  • Join over 1.5M+ people
  • Join over 100K+ communities
  • Free without limits
  • Create your own community
People
Repo info
Activity
  • Aug 01 2018 17:48
    @jnh5y banned @matrixbot
James Hughes
@jnh5y
Does any of Jody's ideas around factories, etc in a JTS 2.0 help with these kinds of issues (again, I'm headed to the deep end)
Martin Davis
@dr-jts
actually you might be on to something there. Having a Z-interpolation strategy in the GeometryFactory could be a good level at which to provide this optionality
Björn Harrtell
@bjornharrtell
Not certain GeometryFactory is the right place.. in either case, I think the approach NTS have taken with a singleton object to provide the global defaults makes sense (but of course it is good if other behaviour can be explicitly chosen at the correct place(s) in the API also)
Martin Davis
@dr-jts
yes, I'm hesitant to load more things onto GeometryFactory. Although it does provide a context that is more fine-grained than global. And having the interpolation associated with the context of a "group" of geometries does make some sense.
Jody Garnett
@jodygarnett
@dr-jts or @jnh5y ping?
Martin Davis
@dr-jts
yessir?
James Hughes
@jnh5y
Yo!
Jody Garnett
@jodygarnett
So a while back eclipse foundation moved to Europe and relaxed a whole bunch.
I asked at the time if we could get our git history back …
… and just go an answer today. Yes we can restore our history (as long as no IP leaking into the main branch).
We are the first project to ask, so Wayne asks that we take some notes. Something like make a branch or tag “pre-eclipse” and then coordinate with “webmaster”.
So yeah, I think we want to do this.
James Hughes
@jnh5y
heheheheh
nice!
Martin Davis
@dr-jts
that would be great!
could switch to main branch at same time perhaps? :)
James Hughes
@jnh5y
sounds like a winning plan!
Martin Davis
@dr-jts
Can we just merge the history from branch _old/history?
Does the major discontinuity due to the change in package name render this less useful?
James Hughes
@jnh5y
Good questions. If we start main from _old/history, that might get us going, so that seems good. It'd depend how the package name changes were handled in Git.
I think Git can either have good info or get confused when files start flying around
Martin Davis
@dr-jts
The package switch too place in _old/history, and it looks to me like git is not able to track files across the boundary. :(
James Hughes
@jnh5y
yeah, I think one may have to have used git mv or something akin to that? (I don't know)
Jody Garnett
@jodygarnett
We may be able to just start from _old/history indeed.
When you check git history you can define a tolerance that helps it track across renames… And git log --follow ./path/to/file can be used as a hint.
Ha ha, lots of folks trying over here.
Björn Harrtell
@bjornharrtell
I still have strong feelings about the regression in locationtech/jts#714 and that it needs to be taken care of before further damage is caused. It could bite anyone who cares about Z-values in a subtle way.. or is it that noone else but me cares about Z? :)
Martin Davis
@dr-jts
You are the only one being vocal about it. :)
I've been thinking about what the least intrusive solution is. Probably the VM flag is simplest.
I would like to make it default to Z-interpolation ON. Does that work for you?
I wonder if you could use Inf as a marker value, and the interpolation code can detect that and avoid interpolation in that case?
Björn Harrtell
@bjornharrtell
Of course turning/forcing it off will work for me but I worry (a little) for others that are less aware. Having inf as magical value doesn't sound good to me, would that not open for ambiguous situations?
James Hughes
@jnh5y
If I recall, JTS's R-Tree implementation is assuming a static dataset. If one needed to add/remove geometries from an R-Tree and wanted to use the R-Tree for knn searches, is an obvious option? (I'm happy to switch to some other data structure as well; no need to stick to an R-Tree:))
Martin Davis
@dr-jts
JTS provides STRtree, which is a semi-static structure. Items cannot be added, but they can be removed.
It's the only index that supports KNN search, at the moment.
Jody Garnett
@jodygarnett
Wow 20 years of JTS!
Jody Garnett
@jodygarnett
I am going to go tweet about that, nothing like thanking Martin in public for being amazing
Martin Davis
@dr-jts
I guess I should make a 20-yr blog post about JTS, to celebrate!
Indeed you should!
Martin Davis
@dr-jts
Thanks, @jodygarnett
Paul did a nice post about PostGIS at 20: http://blog.cleverelephant.ca/2021/05/postgis-20-years.html
Quite a coincidence that JTS and PostGIS were released only 1 day apart, given that we were not working together at all back then
Jody Garnett
@jodygarnett
You young whippersnappers GeoTools has you beat by a few years
Chris Hodgson
@cmhodgson
@dr-jts I believe that "coincidence" is named "Mark" :)
but I guess that is only one piece of the puzzle and still doesn't explain that 1 day apart.
Martin Davis
@dr-jts
@cmhodgson yes, that occurred to me as well. And gov't fiscal year end also plays a part in the alignment.
amittandon
@amittandon
Hey! I wanted to create a grid based clusters (polygons) on a map. I have linesString from different road segments, can anybody guide me how to do it using jts?
Jody Garnett
@jodygarnett
I am not sure I understand what you wish to do, what do you want your polygons to represent? Linestring density or something …