Loading...

Node: 1

addressing warnings in issues tab

Deleted 1 day ago by rjipper
Version #21 · Changeset #81743100
Download XML · View History

see node/1/history

to correct any misunderstanding, osm.org osm-carto style only has a single set of tiles, which make use of the osm.wiki/Tag:`name=' value, and not any osm.wiki/Tag:`name:xx='. although i think what you mean is that the relation name as shown below, which i've never seen on any tiles, is shown for me in english, which is true for my locale. although i wonder how it would be if i were to revert to the `C' locale that i used to use for everything in a past life...

carry on...

salve,

did you mean to add the same name to both segments of this long and winding road, including the still-unnamed way/11113912 ?

thanks

This changeset has been reverted fully or in part by changeset/81792222 where the changeset comment is: delete a maps.me wilderness hut inna big city

This changeset has been reverted fully or in part by changeset/81759373 where the changeset comment is: delete a maps.me mistagged personal home

from what i looked at, it looks as if two mappers are blindly following iDs validator suggestions, doing the usual wikiwhatever tags and apparently a bunch of geometry `fixes'. those changesets refer to solving `issues' and `error nodes' which sounds like blind trust in the iD algorithms, as opposed to specifically joining unconnected ways or whatever.

i've not visualised this changeset or the others; when i did so in Züri some time ago it was obvious that all the changes there were made blindly and wrongly.

while i've not followed the above dg url, i want to point out that this appears to be a mapper working for Apple. there are multiple mappers working for amazon logistics who also have a special licence to use special maxar imagery, presently under clarification in https://forum.openstreetmap.org/viewtopic.php?id=68588 (language german/english).

this is tagged apparently as
imagery_used=Amazon Daily Imagery
source=Latest Private Earth Watch Imagery

and if apple has a similar maxar/dg service to here that is commercially paid for, allowing the results derived from it to appear in osm, then, um, even if normal osm mappers have no direct access to it, it appears to be a source that should be valid for osm, based on the discussion in that forum thread.

please clarify in case i have misrepresented it, thanks

actually, i see two issues here:

amenity waste_basket
combined with
waste dog_excrement
is one thing, as most robidogs and belloos and the like combine a bin together with a dispenser,
amenity vending_machine
fee no
vending excrement_bags

but they are independent functions. near me are dispensers mounted on signposts with no associated bin in the area.

also, while strictly a rabidog is a combo unit intended for both, with a bin for collection of small canine packages, in actual practical use i saw as much regular rubbish being tossed in the bin with the dogpiles, and the collection was the same into a general combined refuse truck also emptying regular bins, which of course may not be true everywhere that separate collection destinations are adhered to...

This changeset has been reverted fully or in part by changeset/81725882 where the changeset comment is: undo commercial seo map vandalism

hallo chepteja and the rest of the amazon team,

this is just general info about how OSM likes to handle old data and accidental deletions like this.

while you have added the driveways deleted, this was very simply done by drawing them from nothing, so that in changeset/81496886 there are links to three ways, but none of them, at v1, will have the original history and credit to the original mapper that can be seen only by following the links below to the deleted way, such as way/756751719/history .

there are tools available that can restore accidentally deleted ways, preserving the history and linking all previous edits.

i know amazon has reported they are planning to migrate editing tools due to deficiencies in the iD editor when deployed on such a large scale; JOSM is often suggested as it has the ability to selectively revert parts of an accidental changeset as this. alternatively, other mappers have access to script tools that can do the same thing, and are willing to apply them in cases like this, as i've just done in a repeat case of deliberate vandalism.

we appreciate you adding back what you deleted, and do want you to know there is a `better' way to do it, that again is a tradeoff between an advanced mapper skill, and what can be expected from entry-level or limited mappers.

thanks

This changeset has been reverted fully or in part by changeset/81538046 where the changeset comment is: repeated vandalism once again, edit wars are fun

agreed, kW and MW are not replacements for kVA and MVA.

I also followed the wiki link to plant:output:electricity=*?uselang=en and there i see there needs to be clarification as to the allowable values (and in the translated languages), and probably an expert should chime in with how kVA and kW differ depending on resistive, reactive, and capacitive loading. i'm no expert, i just pretend to be one when hot-wiring live mains using gaffer tape.

can you be more specific than `paper maps'? most paper maps are under a copyright that prevents their contents from being used for OSM.

also, can you describe what you are doing better than the `vvvvvv' that you have used for multiple edits recently? thanks

you seem to refer to way/776374334 , later referenced by your changeset changeset/81476804 .

this looks to be edited not by an osm editor, but by an old probably customised iD at
created_by iD 2.5.1
hashtags #error;#delete
host http://id.osmsrbija.iz.rs/
which perhaps does not allow one to delete, or something, because normally any editor should allow you to delete a problem way.

just thought i would mention this.

while it's not obvious from the changeset lack of `host' i did click the username to confirm this is an amazon edit. i'm quite sure amazon has special permission or a licence to use imagery that is not generally available to normal osm mappers, and while it has been the subject of discussion for verifiability, i see no problem so long as a more accurate on-the-ground survey would confirm the imagery.

am i trying to say `hands off amazon mappers'? well, you decide. i can't point to the background discussion of this imagery, but it exists, probably in the german forum.

Nachtrag:

in fairness to the issue of iD startup time, i should note that i have observed that it attempts to make a connection to some microsoft mothership, for which i see no reason and which i do not permit, unlike the imagery layers and the like. it could be that iD waits for this connection to fail/succeed; on the other hand i see hundreds of DNS connections to a server the ISP provides that is unresponsive as at present i am trying to debug the half-dozen problems on the ISP side why their 'net connection Does Not Work. i used to get paid well for this. anyway, if iD depends on being able to talk to microsoft, fakebook, and a plethora of other sites i don't give access to and have seen that iD functionality breaks, well, my observations are probably not relevant. just setting the record straight, i hope.

dunno why, but the obsessive-compulsive in me has decided to look more into this behaviour, and thanks for your observations.

i started an iD window - it took several minutes to get beyond the white screen, but eventually it did. admittedly i've got far too many tabs open, three with dubious javascript that looks to be a memory hog to monitor my public services.

partial output of cpu-sorted `top':
12324 freebeer 20 0 1754892 787984 10384 S 10,6 23,7 513:35.32 plugin-con+

(this is a separate process for multiple potlatch 1 and 2 windows with work-in-progress, including a pulled-and-joined node that p1 can't revert the way i tried before temporarily giving up)

21253 freebeer 20 0 457388 8652 1844 S 5,3 0,3 2:04.42 avplay
2227 root 20 0 328688 58244 28124 S 3,3 1,7 390:46.04 Xorg
11953 freebeer 20 0 3412580 1,988g 54388 S 1,3 62,6 3845:16 firefox

i really need to close multiple tabs. last time i paid attention the main browser process was around 30% memory, now there is discernible thrashing, and like i said, i was ready to give up on starting iD with no visible action for several minutes.

2237 freebeer 20 0 42528 464 0 S 0,3 0,0 15:23.70 twm

call me old-fashioned, but it works great for me.

anyway, while my potlatch tabs are all in a separate process, the iD thread-or-whatever is part of the monolithic firefox and has put its memory use from 50-something to over 60%.

when i opened osm.org/edit?node=1036681089#map=19/42.31805/-72.64060 in potlatch1, it took longer than normal but not so long as this iD session. so much for a lightweight simple editor for beginners. to make things worse, i have a pile of Handys older than this machine that are even more memory-limited, where the browser will be killed within seconds of starting iD in order to visualise a changeset.

anyway...

i've edited road 400 in iD, deliberately did not save, and closed the window. during this process i could not identify where iD checkpointed the work - `lsof' showed nothing, nor did the usual suspects like /tmp or ~/ reveal anything. i think i need to dig in the source code rather than dissect my filesystem, the goal being to see what and how is being checkpointed, and where.

regards the small changesets, there are a few issues i saw, https://github.com/openstreetmap/iD/issues/1598 and https://github.com/openstreetmap/iD/issues/2251 and probably https://github.com/openstreetmap/iD/issues/703 which disappeared from my `lynx' browser history but not me brane so if the link is wrong, blame me.

oh blast, i meant to try to start two simultaneous iD instances. but since i haven't found where to look yet, not sure if that would get me anywhere.

anyway, i don't have much to offer so i may as well mash the `Comment' button and go to sleep.

+100 for opening hours. it's been an eternity since i've seen a bancomat that is only available when the bank is open, as the banks cut back on their hours to save money.

further, with, say, a wall of bancomats within a bank building, there may be separate functions that have been proposed for tagging. one may accept payments in, another may give the full range of Euro notes from 5 to 500 (minus the 200) at the time I saw it, another may give swiss francs and a limited selection of Euroen, all of which tags are irrelevant to the bank proper but can be very important to the bancomat customer seeking a particular function or international network.

yeah, now that i look at this, the v1 is a clone of the edited v6 (names need tweaking in indiana, some of the roads with this naming pattern have been expanded to South but not this S), it is not within the bbox of this edit which should be limited to the single car park service drive in pennsylvania. there's probably some explanation why colorado is involved.

the two edits linked in
<< 81158988 · ssange · 81159095 >>
before and after the v6 edit give no clue how the road got cloned about 1000 changesets total later. here we have
<< 81159884 · ssange · 81160031 >>
in michigan and minnesota, so no clue why colorado should be involved at all.

all these edits were made via the osm website, while i still see other amazon edits without the given host (internal amazon iD instance). there is a common bug where amazon editors save multiple distinct edits to a single save creating large changeset bboxen; today i have suggested opening a new broswer tab for each edit rather than re-using an older edit tab when the editing session appears to freeze.

however, i recall reading somewhere that iD checkpoints unsaved data, so this might not be such a good idea to be running multiple editor instances, which i routinely due with potlatch which does not checkpoint unsaved data anywhere i know of, meaning unsaved work gets lost when the oom-killer reaps the flash instances. so it may well be that my suggestion of multiple tabs may make things worse, with each iD instance claiming the unsaved data for itself.

obviously my earlier suggestion was not well thought out, and won't help with the known problems, so i should go back to my day job hauling stones and leave anything requiring thought and intelligence to the experts.

as for identifying edits that cover too large an area, once an edit has been uploaded and saved, you cannot change its area.

as far as i know, there is nothing to change after the fact; the two edits that form the bounding box visualised in the changeset overview were probably correct and the only thing to change is to split them into separate edits by saving and waiting for the save to complete before going on to edit elsewhere, or invoking a new instance of iD in a new tab or window, with the overhead and the time which on my machine seems to border on an eternity, whilst Potlatch 1 is up nearly as fast as I can click through the startup (while being not suited for general editing these days).

i don't know what sort of correction your team is considering after the fact - a revert is possible but also covers the same large area with the problems reflected in the original, making a second changeset covering multiple us states to appear in the history tools, and bringing no benefit as both sets of edits would need to be repeated, losing time.

i can't see how identifying and correcting the problem changesets could help, but perhaps i'm not thinking outside the bbox enough, so i (and probably others) would like to hear details about this proposed course of action.