<html theme="default-light" iconset="color"><head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head><body text="#000000">Taking it a bit off topic.<br>
<br>
Size makes a difference. A relationship of two nodes that no one
touches very much doesn't matter very much on the performance side. A
relationship with lots of nodes or ways or whatever is more likely to
have something changed. When a change occurs, and I'm guessing at this
point based on experience we write the entire relationship to the
history database.<br>
<br>
Then we get into does the history database matter. Well it's useful to
look at the mapper sometimes, one who has mapped twice three years ago I
might just correct something. One with 20,000 edits and mapped
yesterday I might try a changeset comment. I think Fredrick commented
that the history file for some relations is of such a size that it times
out before being able to return the history. It does depend on
popularity. One local foot bridge next to Ottawa University seems to
attract lots of students who correct it by an inch or so almost like a
rite of passage.<br>
<br>
The other side of keeping databases slim and tidy might be to assume
every highway in Africa apart from residential has a default value of
unlit. We don't make that assumption or anything similar but there are a
number of highways in remote parts that are tagged lit=no. Having an
unstated default value is not uncommon in databases, it keeps things
mall but the value can be calculated. I don't expect this will ever
happen in OSM the decision making to too diverse.<br>
<br>
The final comment would be not every mapper knows how to change
something in a relationship. The problem here becomes were someone
wants to correct the local map but is unable to do so without spending
time working out how to do it. Locally we have one or two mappers with
very specialist knowledge who don't map in OSM very much but their
contributions are accurate.<br>
<br>
Cheerio John<br>
<br>
<span>Jarek Piórkowski wrote on 10/20/2021 9:40 PM:</span><br>
<blockquote type="cite"
cite="mid:CACV3h2kr-a=n2U-7CBOy5bwnwDx8fZnB+gnWtV04WLZZeqmSkg@mail.gmail.com">
<pre wrap="">On Wed, 20 Oct 2021 at 17:15, Frederik Ramm <a class="moz-txt-link-rfc2396E" href="mailto:frederik@remote.org"><frederik@remote.org></a> wrote:
</pre>
<blockquote type="cite"><pre wrap="">What's more, these waterbodies do not have an observable or even well
defined outer boundary, forcing waterbody mappers to invent random
straight lines on the far side of some gulf or bay or whatever. This
runs counter to our maxim of mapping what is verifiable on the ground.
</pre></blockquote>
<pre wrap="">
This is true of most natural features. When will we be deleting the
Alps (relation/2698607) or the Berliner Urstromtal (relation/2218270)?
If I wanted to push things to an absurd extreme, coasts are not
observable on a micro scale, clearly anyone mapping coasts is doing so
along a made-up boundary just to get a nice blue body of water roughly
where they think it should be.
I would also be interested as to how political boundaries (of
municipalities, states, etc) are verifiable on the ground their entire
length. We can start keeping the database slim and tidy by deleting
those.
--Jarek
_______________________________________________
Talk-ca mailing list
<a class="moz-txt-link-abbreviated" href="mailto:Talk-ca@openstreetmap.org">Talk-ca@openstreetmap.org</a>
<a class="moz-txt-link-freetext" href="https://lists.openstreetmap.org/listinfo/talk-ca">https://lists.openstreetmap.org/listinfo/talk-ca</a>
</pre>
</blockquote>
<br>
<div class="moz-signature">-- <br>
<div>Sent from <a href="https://www.postbox-inc.com"><span style="color:
rgb(0, 157, 247);">Postbox</span></a></div></div>
</body></html>