[OSM-talk] How to teach novices about optimal changeset size?

[Michael Collinson]

Hi Micah,

I think you came up with a good answer to your conundrum in an earlier 
post in this thread: Don't explain what an optimal changeset IS, explain 
what it is NOT:

Something like:

"It helps other contributors understand your edits if you group what you 
are doing in a local area into one changeset. For example, if you are 
creating the outlines of 20 buildings, group them into one changeset. On 
the other hand, if you are adding 3 POIs, (points of interest),  that 
are 1000 km apart in different countries, then it is more useful to put 
them into 3 changesets. Of course, if you are creating the outlines of 
1,000 buildings in your town, you do not have to do them all at once!

If you worried about losing your data, our data editor software allows 
you to make incremental saves to the OSM server as you go along. iD does 
this automatically. Potlatch and JOSM have buttons that allow you to 
save partial work into a changeset and then keep adding to it until you 
are done."

[This could probably be improved for readability by non-native English 
speakers. And the editor text should be fact checked, I am a die-hard 
Potlatch user.]


Mike

(first post for a long, long time)


On 1/17/18 4:13 PM, Micah Brzozowski wrote:
> Certainly I am not intending to change the community and require every 
> mapper to comply. If you're an experienced mapper, you're fine.
>
> I mean new users, who are not yet integrated with the community. Their 
> work should be checked thoroughly (in Achavi, osmcha...). All novices 
> make mistakes, after all. Better to give them good habits. By 
> extension, smaller number of changeset will lead to less recycling of 
> same changeset comments.
>
> I made this thread because I found it difficult to convey what is best 
> practice in short form in changeset comments.
>
> Maybe I should simplify things when explaining to them? No need to 
> tell all the conventions, just what is a good start - but hoping it 
> won't backfire ;)
>
> 17.01.2018 3:35 PM "Imre Samu" <pella.samu at gmail.com 
> <mailto:pella.samu at gmail.com>> napisał(a):
>
>     >  one changeset per building, repeated 20 times
>
>     my typical use case:   House numbering on the street: push the
>     numbers & forget & go to the next house    ( fast feedback loop
>     vs. Delayed gratification  )
>     - sometimes the mobil app is crashing, and I don't want to go back
>     100m to re-enter - the last 5-10 numbers
>
>
>     > Obviously this makes them PITA to review quickly in Achavi or
>     whatever tool you use.
>
>     imho: it is easier to group the changeset on the reviewer side : 
>     by user + by hour   ( group by user, hour )   than change the
>     community.
>
>     Imre
>
>
>
>
>
>     2018-01-17 15:13 GMT+01:00 Michał Brzozowski <www.haxor at gmail.com
>     <mailto:www.haxor at gmail.com>>:
>
>         Certainly not:
>         - one changeset per building, repeated 20 times
>         - one changeset for 3 POIs that are 1000 km apart in different
>         countries
>
>         These are real world examples. In the latter Achavi can often
>         refuse to run.
>
>         That's also why I asked ;-) It's not that easy to formulate
>         the answer what is reasonable to include in a changeset.
>
>         Michał
>
>         17.01.2018 2:54 PM "Tobias Zwick" <osm at westnordost.de
>         <mailto:osm at westnordost.de>> napisał(a):
>
>             So, what is the optimal changeset size, and why?
>
>             Tobias
>
>             On 17/01/2018 14:26, Michał Brzozowski wrote:
>             > Many new users have a habit of e.g. sending one or few
>             objects per
>             > changeset, resulting in a dozen or even more changesets
>             per day.
>             > Obviously this makes them PITA to review quickly in
>             Achavi or whatever
>             > tool you use.
>             >
>             > This habit is probably caused by non-knowledge of how
>             auto-save works in
>             > iD (which makes the work reasonably secure), as well as
>             just not knowing
>             > better thus forming their own judgement.
>             >
>             > How should we teach about optimal changeset size? This
>             is quite tricky -
>             > how we would define it?
>             >
>             > Can the iD nudge users towards better practice? (Linking
>             to Good
>             > changeset comments wiki page would be useful as well)
>             >
>             > Michał
>             >
>             >
>             > _______________________________________________
>             > talk mailing list
>             > talk at openstreetmap.org <mailto:talk at openstreetmap.org>
>             > https://lists.openstreetmap.org/listinfo/talk
>             <https://lists.openstreetmap.org/listinfo/talk>
>             >
>
>
>             _______________________________________________
>             talk mailing list
>             talk at openstreetmap.org <mailto:talk at openstreetmap.org>
>             https://lists.openstreetmap.org/listinfo/talk
>             <https://lists.openstreetmap.org/listinfo/talk>
>
>
>         _______________________________________________
>         talk mailing list
>         talk at openstreetmap.org <mailto:talk at openstreetmap.org>
>         https://lists.openstreetmap.org/listinfo/talk
>         <https://lists.openstreetmap.org/listinfo/talk>
>
>
>
>
>
> _______________________________________________
> talk mailing list
> talk at openstreetmap.org
> https://lists.openstreetmap.org/listinfo/talk

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/talk/attachments/20180119/e964d31c/attachment-0001.html>