New ways not displayed

[stoecker]

Under rare conditions new ways are not displayed in JOSM. Zooming a bit in and out usually shows them again. I was not yet able to make it reproducable. Probably an issue with wrong handling in QuadBuckets.

[Gubaer]

This happens very often while mapping. I'll collect sample bounding boxes with sample objects to reproduce and I'll post them here.

[vsandre]

I have downloaded a small area and than all members of a containing route relation. And now the problem appears, no ways were shown. Afterwards I saved and deleted the data layer, with the aim to reproduce the bug, but all works fine. Also the saved file looks fine.

This bug is very annoying and could leads into wrong edits.

[vsandre]

BTW selecting the route relation shows all ways.

[Ldp]

Example of this happening for me

[Ldp]

Load my example file, and drag one of the nodes. The way disappears, while the nodes are still painted. Zoom out pretty far to get the way painted again.

[Ldp]

I went back to r2000 and worked my way back in steps, and I found my use case (see disappearing_ways.osm) was introduced in r2450 ('Added parameter Bounds to MapView, draw only currently visible primitives in MapPaintVisititor').

I had personally classified this (disappearing ways) as a Heisenbug, because I could never reproduce it before.

[malenki]

This bug could be related to #4111. Trying to reproduce it with r2864 did not result in a disappearing way but in a way who couldn't be selected anymore. Nodes can be selected, dragged around and used to continue the way. The way itself remains unselectable.

[jttt]

I can confirm that with disappearing_ways.osm. I'm working on a fix.

[jttt]

(In [2868]) Fixed #4324 New ways not displayed

[malenki]

In r2887 I experience a not straightforward reproducible error in wireframe view.

It appeared as I worked on Nile with Lakewalker. Sometimes newly created ways weren't displayed at all (only the nodes), sometimes parts of the way vanished after splitting. See attachments

[malenki]

new_way_complete

[malenki]

new_way_complete

[malenki]

way_splitted_disappeared

[malenki]

way_still_exists

[malenki]

PS: I forgot to overwrite "gkrellShoot_2010-01-24_134508_new_way_complete.png" - can be deleted (or just disregard it)

[malenki]

Changed to blocker since though the ways are diplayed in mappaint they don't exist. Validator sais there are thousands of empty nodes. Ways which weren't changed but happened to be displayed also disappeared leaving only the nodes behind. See attachment. Some hours of work wasted :€

[malenki]

For the attachment nil_broken.osm: If it can be repaired easily it would nice if it would be uploaded (without the deletions at HammerHeads ways and nodes). Would hate if I really had wasted so much time...

[jttt]

What Dataset consistency test says? (right click on osm layer)

[malenki]

Hrm. This i forgot to perform as I reported the bug. The Joms instance where this happened is long since closed. To the attached file the consistency test sais:

{Way id=5229400 version=6 V nodes=[]} not found using Dataset.searchWays()
Way {Way id=5229400 version=6 V nodes=[]} has zero nodes

[jttt]

Hm, next time this kind of bug happens make also heap dump (see #4057 for details).

[jttt]

Do you remember when the problems started? From the beginning, after data updated or some other event?

Also do I understand it right that the ways were displayed in mappaint mode, but not in wireframe mode?

[jttt]

It looks like #4057 strikes again. And again lakewalker was used.

[anonymous]

Replying to jttt:

It looks like #4057 strikes again. And again lakewalker was used.

If you think this is true, it's not true in every case. I have not used lakewalker, and don't even have it loaded, and yet (as can be seen in disappearing_ways.osm) I also have intermittent issues. Split in 2 tickets?

[Ldp]

Replying to anonymous:

Replying to jttt:

It looks like #4057 strikes again. And again lakewalker was used.

If you think this is true, it's not true in every case. I have not used lakewalker, and don't even have it loaded, and yet (as can be seen in disappearing_ways.osm) I also have intermittent issues. Split in 2 tickets?

Anonymous replies sometimes considered harmful. This was me.

[jttt]

The original issue is fixed in r2868 (the one with disappearing_ways.osm), the problem reported by malenki is different (and yes, it should have been in separated ticket).

[malenki]

Replying to jttt:

Do you remember when the problems started? From the beginning,

no, just some time after I started editing

Also do I understand it right that the ways were displayed in mappaint mode, but not in wireframe mode?

yes

[malenki]

Replying to Ldp:

Replying to anonymous:

Replying to jttt:

It looks like #4057 strikes again. And again lakewalker was used.

yes (though I didn't mention it explicitly)

If you think this is true, it's not true in every case. I have not used lakewalker, and don't even have it loaded, and yet (as can be seen in disappearing_ways.osm)

The attachment contains broken data so it is not relevant imo which plugins you at your josm are active.

[malenki]

Replying to jttt:

The original issue is fixed in r2868 (the one with disappearing_ways.osm) [was created, you mean?]

the problem reported by malenki is different

How should I have known?

(and yes, it should have been in separated ticket).

Next time one has to tell me before reopening/adding a a ticket. One time I was "scolded" not having reopened a ticket, now I am having done so. Sums up to zero :)

[stoecker]

Only use existing tickets when you are sure the issue is the same. Use a new one otherwise. You get blamed correctly when you open a new ticket and the same issue has been reported shortly before in another one. Ignore any blaming when the previous ticket is older or could not be found with an easy search.

It is not the task of a user to maintain the tickets or guess whether they really represent equal topics, but he should also not increase developers load by obvious duplicates.

[jttt]

I might have fixed the bug, see #4057 for details.

[anonymous]

Closed as duplicate of #4057.