<div dir="auto">That is incorrect, some building parts could be bigger if they are surrounding the building as an overhang etc. You can't assume building will be bigger</div><br><div class="gmail_quote"><div dir="ltr">On Thu., Jan. 24, 2019, 11:51 a.m. Nate Wessel <<a href="mailto:bike756@gmail.com">bike756@gmail.com</a> wrote:<br></div><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex">
  
    
  
  <div text="#000000" bgcolor="#FFFFFF">
    <p>Is it sufficient to tag fragments as building:part without
      indicating which part or how many stories? If the data is properly
      structured, this seems like something that could be handled in
      preprocessing by checking for overlapping polygons. It looks like
      perhaps we might just have to find the largest part for the
      footprint (building=yes) and any intersecting smaller buildings
      (building:part=yes).</p>
    <p> We might also need to generate some building relations for more
      complex features.<br>
    </p>
    <div class="m_-6862194445823564595moz-signature">Nate Wessel<br>
      <span style="font-size:10px;color:#777">Jack of all trades, Master
        of Geography, PhD candidate in Urban Planning<br>
        <a href="http://natewessel.com" target="_blank" rel="noreferrer">NateWessel.com</a></span>
      <br>
      <br>
    </div>
    <div class="m_-6862194445823564595moz-cite-prefix">On 1/24/19 11:40 AM, Yaro Shkvorets
      wrote:<br>
    </div>
    <blockquote type="cite">
      
      <div dir="ltr">
        <div dir="ltr">
          <div dir="ltr">
            <div>OSM wiki: <a href="https://wiki.openstreetmap.org/wiki/Key:building:part" target="_blank" rel="noreferrer">https://wiki.openstreetmap.org/wiki/Key:building:part</a></div>
            <div>It's not in the import wiki though since whoever wrote
              it didn't know about it at the time.</div>
            <div>Here's what I mean by mapping 3D features in our case.
              Say there is a residential tower on a podium. In the
              StatsCan data usually you would find both of these
              outlines - large podium outline and smaller tower outline
              inside it. They would both be tagged with "building=yes"
              tag. Obviously we can't upload that as-is. We can either
              just remove tower outline leaving only 2D podium outline.
              Or, we can tag the tower outline with "building:part=yes".
              Someone local can add other tags to it later on, such as
              "building:levels", "building:material",
              "building:min_level", "addr:housenumber" (if there are two
              towers on one podium with different house numbers for
              example), etc. I find the latter approach to be the right
              one.</div>
          </div>
        </div>
      </div>
      <br>
      <div class="gmail_quote">
        <div dir="ltr" class="m_-6862194445823564595gmail_attr">On Thu, Jan 24, 2019 at 11:15
          AM Nate Wessel <<a href="mailto:bike756@gmail.com" target="_blank" rel="noreferrer">bike756@gmail.com</a>> wrote:<br>
        </div>
        <blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">
          <div bgcolor="#FFFFFF">
            <p>Hi Yaro, <br>
            </p>
            <p>I just had a chance to look at the documentation on the
              source data and I wasn't able to find anything about 3D
              features or parts of buildings being mapped separately.
              Are you guessing here, or is there documentation on this?
              If so can you point us to it?<br>
            </p>
            <p>In any case, the big shapefiles from StatsCan don't
              provide enough information to reconstruct any 3D
              geometries, so I'd be inclined to remove these from the
              import unless they can be brought in from another source
              with better documentation / attribute tagging. (i.e. City
              of Toronto?)<br>
            </p>
            <p>Thanks,<br>
            </p>
            <div class="m_-6862194445823564595gmail-m_277083992809862550moz-signature">Nate
              Wessel<br>
              <span style="font-size:10px;color:rgb(119,119,119)">Jack
                of all trades, Master of Geography, PhD candidate in
                Urban Planning<br>
                <a href="http://natewessel.com" target="_blank" rel="noreferrer">NateWessel.com</a></span> <br>
              <br>
            </div>
            <div class="m_-6862194445823564595gmail-m_277083992809862550moz-cite-prefix">On
              1/18/19 2:48 PM, Yaro Shkvorets wrote:<br>
            </div>
            <blockquote type="cite">
              <div dir="ltr">
                <div dir="ltr">Jarek,
                  <div>There is no question we want this data. I went
                    through much of it in Toronto and Kingston and I
                    found it to be very good, consistent and precise.
                    Time-wise it's somewhat current with 2016 ESRI
                    imagery (sometimes ahead, sometimes slightly behind)
                    and is well-aligned with it. It offers 3D features
                    (when several buildings appear overlapped in the
                    dataset) but you just need to be familiar with
                    `building:part` tag to sort through it. I haven't
                    looked at other provinces but in Ontario I really
                    have no complaints about dataset quality whatsoever.
                    Also I don't get Nate's "wildly unsimplified
                    geometries" comment. IMO geometries are just
                    perfectly detailed.</div>
                  <div><br>
                  </div>
                </div>
              </div>
              <br>
              <div class="gmail_quote">
                <div dir="ltr">On Fri, Jan 18, 2019 at 2:00 PM Jarek
                  Piórkowski <<a href="mailto:jarek@piorkowski.ca" target="_blank" rel="noreferrer">jarek@piorkowski.ca</a>>
                  wrote:<br>
                </div>
                <blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex">Some more thoughts
                  from me.<br>
                  <br>
                  Building outlines, particularly for single-family
                  subdivisions as seen<br>
                  in Canadian suburbs, are extremely labour-intensive to
                  map manually.<br>
                  <br>
                  My parents' house is now on OSM - accurately. They
                  live in a city with<br>
                  about 10,000 buildings, and about 0.5 active mappers.
                  This wouldn't<br>
                  been completed manually in the next 5 years.<br>
                  <br>
                  An option to do this automatically with a computer
                  algorithm detecting<br>
                  objects from imagery could be suggested, but this has
                  not been very<br>
                  accurate in OSM in the past, even when there is decent
                  imagery. The<br>
                  only other feasible data source is government, where
                  they have such<br>
                  data more or less.<br>
                  <br>
                  The alternative is of course the opinion that we
                  should not have<br>
                  building outlines until someone goes through and adds
                  the buildings<br>
                  manually. In practice what I've seen done in Toronto
                  is that bigger<br>
                  buildings are mapped on best-effort basis from survey
                  and imagery,<br>
                  while areas of single-family houses are left blank.
                  This isn't<br>
                  _wrong_, and maybe some prefer this.<br>
                  <br>
                  I would also like to note that building outlines will
                  _never_ be<br>
                  completely verifiably up to date. I can't go into most
                  people's<br>
                  backyards and verify that there isn't a new addition
                  on their house. A<br>
                  building might be legally split into two different
                  properties without<br>
                  it being evident from the street. Imagery is out of
                  date the day after<br>
                  it's taken, and proper offset can be difficult to
                  establish in big<br>
                  cities where GPS signal is erratic. Pragmatically, I
                  can tell you from<br>
                  personal experience that building data in
                  lovingly-mapped Berlin is<br>
                  also worse than 1 meter accuracy. So again: best
                  effort.<br>
                  <br>
                  What do we get from having buildings? A sense of land
                  use (arguably<br>
                  replaceable with larger landuse areas). A way to
                  roughly estimate<br>
                  population density. A way to gauge built-up density. A
                  data source for<br>
                  locating buildings in possible flood zones, or fire
                  risk. Statistics:<br>
                  as open data, queryable by APIs that are already used,
                  in format<br>
                  more-or-less common worldwide.<br>
                  <br>
                  Examples were given of rowhouse- or de-facto
                  rowhouse-buildings where<br>
                  a part is attached to the wrong building. This does
                  not alter any of<br>
                  the above examples. It's wrong, but is it
                  substantially more wrong<br>
                  than a blank subdivision, or one with only a few
                  buildings mapped? Is<br>
                  it better to have a null, or be off by 5%? The legal
                  truth is in<br>
                  property records, and we can't measure houses with a
                  ruler, so OSM can<br>
                  only be a statistical source. And then there's the
                  question of<br>
                  verifiability - some of these buildings are connected
                  to their<br>
                  neighbour building inside. I've really struggled at
                  distinguishing<br>
                  what exactly is a "building" on Old Toronto avenues
                  even with<br>
                  street-side survey.<br>
                  <br>
                  Bluntly, OSM is not perfect in Canada. I have pet
                  peeves I can quote,<br>
                  and I'm sure many of you do as well. If we import, the
                  question is:<br>
                  are we making it better?<br>
                  <br>
                  1. Do we want this data?<br>
                  2. Is it generally of acceptable quality?<br>
                  3. Is there a mechanism to spot and reject where data
                  is particularly bad?<br>
                  <br>
                  Cheers,<br>
                  Jarek, who should really get back to updating
                  built-last-year stuff at Fort York<br>
                  <br>
                  On Fri, 18 Jan 2019 at 09:31, Kyle Nuttall <<a href="mailto:kyle.nuttall@hotmail.ca" target="_blank" rel="noreferrer">kyle.nuttall@hotmail.ca</a>>
                  wrote:<br>
                  ><br>
                  > The pilot project that took place in Ottawa for
                  all these building imports is what got me hooked into
                  OSM in the first place. I would make only very minor
                  changes here and there. I even attempted to draw
                  building footprints but got burnt out after only doing
                  a single street, which was very discouraging for me to
                  continue.<br>
                  ><br>
                  > When I saw the entire neighbourhood get flooded
                  with new buildings that weren't there before, I was
                  entirely intrigued and actually got on board with the
                  locals to help with the process. I've been hooked
                  since and have been to many meetups afterwards.
                  Helping out with projects completely unrelated to the
                  initial building import.<br>
                  ><br>
                  > I'm entirely of the belief that it is much more
                  encouraging for a new user to make a minor change (eg.
                  changing `building=yes` to `building=detached`) than
                  it is to add every single minor detail to each object
                  from scratch (visiting the location, drawing the
                  building footprints manually, adding address data,
                  etc.). It's just overwhelming for a new user.<br>
                  ><br>
                  > It is very much a cat-and-mouse type scenario
                  with community driven projects like OSM. Apparently
                  the issue with this import is the lack of community
                  involvement but I can for sure tell you that this
                  import will help flourish the community in the local
                  areas. Especially if they only need to add or change
                  minor tags than if they would have had to create all
                  of this data by hand. With an import this size there
                  is bound to be some errors that slip through. That's
                  where the community comes through to correct these
                  minor things.<br>
                  ><br>
                  > This is the whole point of OSM. A user creates an
                  object with as much information as they know and the
                  next user comes and adds onto that, and the next user
                  adds and/or updates even more. Neither of those users
                  on their own could have added as much detail as all of
                  their knowledge combined.<br>
                  ><br>
                  > Are we supposed to just wait for a user who can
                  add every single building with centimetre precision
                  and every bit of detail simply because we can't? No,
                  of course not. We do the best we can and have other
                  users who know more than we do build on that.<br>
                  ><br>
                  > I fully endorse this import because I would love
                  to see what it does for the local communities that
                  apparently need to figure this import out for
                  themselves.<br>
                  ><br>
                  > Cheers,<br>
                  > Kyle<br>
                  ><br>
                  > On Jan. 18, 2019 05:40, James <<a href="mailto:james2432@gmail.com" target="_blank" rel="noreferrer">james2432@gmail.com</a>>
                  wrote:<br>
                  ><br>
                  > As Frederik Ramm once said(sorry i'm paraphrasing
                  from memory please don't shoot me) There has never
                  been a GO-Nogo for imports, you bring it up on the
                  mailing lists with reasonable delay, is there no
                  objections(in this case no one was saying anything
                  about it for 2-3 weeks) then email the list that the
                  import would start.<br>
                  ><br>
                  > On Fri., Jan. 18, 2019, 12:59 a.m. Alan Richards
                  <<a href="mailto:alarobric@gmail.com" target="_blank" rel="noreferrer">alarobric@gmail.com</a>
                  wrote:<br>
                  ><br>
                  > Along the lines of what Jarek said, sometimes
                  silence just means tacit acceptance, or that it's not
                  that controversial. There's quite a bit of government
                  data here that is supposedly "open" but unavailable
                  for OSM, so I'm very glad Stats Can was able to find a
                  way to collect municipal data and publish it under one
                  national license. I was surprised myself it hadn't got
                  more attention, but I'm firmly onboard with more
                  imports if done with care.<br>
                  > Manually adding buildings - especially
                  residential neighborhoods, is about the most boring
                  task I can think of, yet it does add a lot to the map.<br>
                  ><br>
                  > I'll admit I hadn't looked at the data quality
                  myself, but I just did review several task squares
                  around BC and they look pretty good. Houses were all
                  in the right place, accurate, and generally as much or
                  even more detailed than I typically see. Issues seemed
                  to be mostly the larger commercial buildings being
                  overly large or missing detail, but in general these
                  are the buildings most likely to be already mapped. To
                  a large degree, it's up the individual importer to do
                  some quality control, review against existing object,
                  satellite, etc. If we have specific issues we can and
                  should address them, but if the data is largely good
                  then I see no need to abort or revert.<br>
                  ><br>
                  > alarobric<br>
                  ><br>
                  > On Thu, Jan 17, 2019 at 7:41 PM Jarek Piórkowski
                  <<a href="mailto:jarek@piorkowski.ca" target="_blank" rel="noreferrer">jarek@piorkowski.ca</a>>
                  wrote:<br>
                  ><br>
                  > On Thu, 17 Jan 2019 at 21:46, OSM Volunteer
                  stevea<br>
                  > <<a href="mailto:steveaOSM@softworkers.com" target="_blank" rel="noreferrer">steveaOSM@softworkers.com</a>>
                  wrote:<br>
                  > > Thanks, Jarek.  Considering I am a proponent
                  of "perfection must not be the enemy of good"
                  (regarding OSM data entry), I think data which are
                  "darn good, though not perfect" DO deserve to enter
                  into OSM.  Sometimes "darn good" might be 85%, 95%
                  "good," as then we'll get it to 99% and then 100% over
                  time.  But if the focus on "how" isn't sharp enough to
                  get it to 85% (or so) during initial entry, go back
                  and start over to get that number up.  85% sounds
                  arbitrary, I know, but think of it as "a solid B"
                  which might be "passes the class for now" without
                  failing.  And it's good we develop a "meanwhile
                  strategy" to take it to 99% and then 100% in the
                  (near- or at most mid-term) future.  This isn't
                  outrageously difficult, though it does take patience
                  and coordination.  Open communication is a
                  prerequisite.<br>
                  ><br>
                  > Thank you for this commitment. I wish others
                  shared it. Unfortunately<br>
                  > the reality I've been seeing in OSM is that edits
                  which are 90+% good<br>
                  > (like this import) are challenged, while edits
                  which are 50+% bad<br>
                  > (<a href="http://maps.me" rel="noreferrer noreferrer" target="_blank">maps.me</a>
                  submissions, wheelmap/rosemary v0.4.4 going to
                  completely<br>
                  > wrong locations for _years_) go unchallenged or
                  are laboriously<br>
                  > manually fixed afterward.<br>
                  ><br>
                  > --Jarek<br>
                  ><br>
                  > _______________________________________________<br>
                  > Talk-ca mailing list<br>
                  > <a href="mailto:Talk-ca@openstreetmap.org" target="_blank" rel="noreferrer">Talk-ca@openstreetmap.org</a><br>
                  > <a href="https://lists.openstreetmap.org/listinfo/talk-ca" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/talk-ca</a><br>
                  ><br>
                  > _______________________________________________<br>
                  > Talk-ca mailing list<br>
                  > <a href="mailto:Talk-ca@openstreetmap.org" target="_blank" rel="noreferrer">Talk-ca@openstreetmap.org</a><br>
                  > <a href="https://lists.openstreetmap.org/listinfo/talk-ca" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/talk-ca</a><br>
                  ><br>
                  ><br>
                  > _______________________________________________<br>
                  > Talk-ca mailing list<br>
                  > <a href="mailto:Talk-ca@openstreetmap.org" target="_blank" rel="noreferrer">Talk-ca@openstreetmap.org</a><br>
                  > <a href="https://lists.openstreetmap.org/listinfo/talk-ca" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/talk-ca</a><br>
                  <br>
                  _______________________________________________<br>
                  Talk-ca mailing list<br>
                  <a href="mailto:Talk-ca@openstreetmap.org" target="_blank" rel="noreferrer">Talk-ca@openstreetmap.org</a><br>
                  <a href="https://lists.openstreetmap.org/listinfo/talk-ca" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/talk-ca</a><br>
                </blockquote>
              </div>
              <br clear="all">
              <div><br>
              </div>
              -- <br>
              <div dir="ltr" class="m_-6862194445823564595gmail-m_277083992809862550gmail_signature">
                <div dir="ltr">Best Regards,<br>
                            Yaro Shkvorets</div>
              </div>
              <br>
              <fieldset class="m_-6862194445823564595gmail-m_277083992809862550mimeAttachmentHeader"></fieldset>
              <pre class="m_-6862194445823564595gmail-m_277083992809862550moz-quote-pre">_______________________________________________
Talk-ca mailing list
<a class="m_-6862194445823564595gmail-m_277083992809862550moz-txt-link-abbreviated" href="mailto:Talk-ca@openstreetmap.org" target="_blank" rel="noreferrer">Talk-ca@openstreetmap.org</a>
<a class="m_-6862194445823564595gmail-m_277083992809862550moz-txt-link-freetext" href="https://lists.openstreetmap.org/listinfo/talk-ca" target="_blank" rel="noreferrer">https://lists.openstreetmap.org/listinfo/talk-ca</a>
</pre>
            </blockquote>
          </div>
          _______________________________________________<br>
          Talk-ca mailing list<br>
          <a href="mailto:Talk-ca@openstreetmap.org" target="_blank" rel="noreferrer">Talk-ca@openstreetmap.org</a><br>
          <a href="https://lists.openstreetmap.org/listinfo/talk-ca" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/talk-ca</a><br>
        </blockquote>
      </div>
      <br clear="all">
      <div><br>
      </div>
      -- <br>
      <div dir="ltr" class="m_-6862194445823564595gmail_signature">
        <div dir="ltr">Best Regards,<br>
                    Yaro Shkvorets</div>
      </div>
    </blockquote>
  </div>

_______________________________________________<br>
Talk-ca mailing list<br>
<a href="mailto:Talk-ca@openstreetmap.org" target="_blank" rel="noreferrer">Talk-ca@openstreetmap.org</a><br>
<a href="https://lists.openstreetmap.org/listinfo/talk-ca" rel="noreferrer noreferrer" target="_blank">https://lists.openstreetmap.org/listinfo/talk-ca</a><br>
</blockquote></div>