Make iD default editor

[jfirebaugh]

This makes iD the default editor on OSM. It's already the first option in the drop down, but this PR makes iD the editor that users land in when they just click the "Edit" label.

I hope to get this merged, now that initial concerns were addressed with the 1.1. release of iD that was just rolled out on OpenStreetMap.org:

• Improved performance
• Added relation support

[tomhughes]

What is the status with regard to the performance issues that prevented us doing this before (see #262 for the previous history).

[tomhughes]

Initial tests suggest that performance (with Firefox 23) is much better than before. It's still a little slow if you have a lot of data loaded, but that can be improved by zooming in more. Of course this is on a fairly powerful desktop so I don't know what other people's experience will be like.

[tomhughes]

Ignoring performance for a minute I just tried to use iD to add a couple of buildings and adjust a road and well frankly it's buggy as hell and so far I've been unable to complete anything.

So the first attempt ended when I tried to move the end of a road and it wound up glued to my mouse pointer and nothing I did (single click, double click, return, escape, ...) would release it and allow me to carry on.

The second and third attempts ended when the cursor got locked in the move mode (the hand icon) and nothing I did would change it and let me select anything.

At this point, after three attempts I've managed to draw one building and delete two nodes from a road. At least the offline storage of edits makes reloading and continuing easy, but having to do that every 30s is not really a reasonable user experience.

No way is this ready to be the default as it stands.

[lxbarth]

@tomhughes -

Ignoring performance for a minute I just tried to use iD to add a couple of buildings and adjust a road and well frankly it's buggy as hell and so far I've been unable to complete anything.

This is not at all my experience with iD and I wonder what's going on.

Could you report the problems you found together with browser and OS on https://github.com/systemed/iD/issues ?

Let's get any concrete issues addressed and move on to the future. I think there's no disagreement that we want to get off of Potlatch 2.

[tomhughes]

Raised openstreetmap/iD#1691 and openstreetmap/iD#1692 though I'm not sure they're very helpful...

[jfirebaugh]

I believe the issues that @tomhughes encountered in Firefox are fixed in iD 1.1.4. If anyone finds otherwise, please let me know.

[pnorman]

Concerned that doing this before openstreetmap/iD#1646 is fixed will tie the sysadmin's hands for deploying katla/smaug

[pnorman]

Comments expressed on talk@ not already listed here, my notes are in italics. One list item per comment on a topic, per author.

• iD stops propagating non-free insecure software (flash)
• delete function is too prominent for an editor for newcomers
• needs full support for preserving forward/backward/left/right for all keys and values and as role when changing directions and combining ways Exact details required to reproduce or comment on this. Some are handled, see "Reverse Direction" in tag editor should not ignore direction dependent tags iD#299, but it may not of fully implemented JOSM's logic
• possibility to turn off the background Option to disable background entirely iD#1472
• Do not include links to twitter and facebook in an open data project driven by open source applications
• Social features (facebook/twitter links) are good for starting conversations/sharing interests
• Alter workflow for deleting objects Proposed alternative to trash can in radial menu (make "delete" harder to hit). iD#1698
• users can delete relation members without being aware of it Warn before deleting a member of a relation [$15] iD#1461
• Existing turn restrictions need to be preserved Edit turn restrictions iD#224 (comment), new issue created at disable merges that would damage relations iD#1512
• Is having a basic version of iD with opt-in for more advanced features an option?
• A call for details on any non-esoteric relation damage
• immediate warnings when you do something dangerous (and an expert switch to disable them later) would be very helpful to prevent damage and teach the user how to proceed
• Move-around icon on ways but not nodes is confusing to newcomers
• Remove from map confuses map and database
• "click on end node of line and continue drawing it" does not exist or is not obvious
• "copy tags from similar way" does not exist or is not obvious
• Relations are not visible enough when tags push them down to require scrolling
• handling of multipolygons very confusing
• OSM should not create a connection to Facebook, Twitter or any other social service without conscious choice by the user or in a way that suggests that it is an integral part of OSM
• p2 doesn't warn that you're acting destructively" when deleting objects either
• We are trying to replace an aging editor with something more current
• It would be great if we could have some actual data comparing how P2 performs against iD
• Difficult to remember what each button is for without colours

---

Note: Not complete, but heading to bed

[brycenesbitt]

And a few concrete proposals, drawn from discussions:
openstreetmap/iD#1696 (in case iD is too slow on their particular machine/browser combo)
openstreetmap/iD#1698 (change "delete" workflow - move delete button to where the tags are, since we hope users will care to understand an object prior to deleting it. Unclutter POI geometry movement.)
openstreetmap/iD#1706 (make "end of edit" action neutral)

[jfirebaugh]

Thank you for the summary @pnorman. Of the points raised against iD that you listed, the one that is a regression from Potlatch 2 is the lack of a copy/repeat tags feature (P2's 'R' keybinding). We plan to add this to iD, but I would argue that all things considered it's not something that should block a change of default.

In the absence of any other clear bug reports that indicate a regression from Potlatch 2 I think we should merge this and make iD default for new users. I know there's been many good requests made but I don't think we can wait for a hard to define level of perfection to make iD default. The only hard bar we can measure against is the status quo we want to replace. iD is on par with P2 on all critical features and is much easier and safer to use overall.

Let's flip the switch :)

[systemed]

+1.

I'd not only like this to happen because iD is objectively much better for the new user and for many editing cases, but also because it means that, as the maintainer of P2, I can start to gut all the newbie-friendly stuff that P2 has never done very convincingly, and refocus it as the intermediate-level editor for which it's much more suited.

[pnorman]

Could we get a test deployment so that we can test that fallbacks from iD to P2 in browsers iD doesn't support are working correctly?

[tomhughes]

There is only one such fallback and it's pretty trivial so I'm not sure it's worth the effort.

[pnorman]

Okay, as far as I can tell the current regressions are

Technical

• iD should process and apply data returned by the diff upload iD#1646 (parsing diffs, may become an issue in ~1 month)
• I could not find any outstanding performance issues.

User-facing

• disable merges that would damage relations iD#1512 (Relation membership is apparently much less obvious in iD, which is okay, except that if a user is unaware of relation membership they may inadvertently perform merges which damage relations)
• iD endorses proprietary data sites (no issue created yet)

---

openstreetmap/iD#1646 is purely a technical timeline issue. If it gets fixed before we are ready to send map? requests to a slave server, no one will notice. If not, it could blow up badly the first time replication lag exceeds a second or so.

openstreetmap/iD#1512 is on balance, probably not any reason to hold back.

I think more users have expressed opposition to the Twitter and Facebook links than have used the twitter link in the last month.

[systemed]

P2 has a Twitter link now so that's not a reason not to switch (whistles innocently).

[jfirebaugh]

Relation membership is apparently much less obvious in iD

Not "much", and whether or not it's on balance any less obvious is debatable. In some ways they are more visible in P2, in others more visible (and protected) in iD.

I think more users have expressed opposition to the Twitter and Facebook links than have used the twitter link in the last month.

Issue.

By your own count there are about 30 unique users of the Twitter link per month. I count two users opposing any sort of social media sharing entirely and two or three wanting them less prominent or completely hideable.

[tomhughes]

So I really don't see anything here that is a blocker, and performance wise it seems to be OK for me even on my laptop with Firefox 23, so I'm thinking that unless somebody has a very persuasive argument that we haven't already heard we should go ahead and merge this...

[iandees]

👍 Tom. I was just playing around with the latest iD on my older Core 2 Duo Linux laptop and it was noticeably faster than the last time I tried. Very usable.

[emacsen]

+1 the iD

[tomhughes]

Merged.