[Tagging] Deprecation - waterway=riverbank vs water=river

[Marc_marc]

Hello,

Le 10.02.21 à 15:43, Tomas Straupis a écrit :
> safeguards against incompetent

you always have to be careful when you write this kind of judgment.
others could say the same thing about you when you argue that you
shouldn't change A into B because B is less frequent than A
(egg<>chicken), or that you talk about the fact that it might break
something when all serious tools have been handling *both* schemas for
*years*, including rendering style with a "no alias" rule.

> what is next?

my wishes : depreciated any argument based on
- it's incompetence.
- it used to be like that, so it must remain that way
- "too big to fix" dogma, the number of objects that could be
improved is on the contrary an interest to do better, a tag present
once in the base probably does not often need to be discussed here)

going back to our "basic rules of IT":
>   1. Do not touch it if it works.

you don't need to touch any datause, it should already use both tag :)
if you use preset, you don't need to touch anything, preset
do the job.
if you maintain a preset, it require you to change a few words.
to be positive, perhaps we could include that whoever brings
a depreciation proposal, agree to open tickets to all projects
declared via taginfo

>   2. benefits<>cost

if it didn't take "years" to argue against immobility, it would
probably take only a few hour, which is very low compared to the time it
takes to code one more alias for "every" use of the data multiplied by
the number of times it happens (for overpass requests alone,
it's very boring to have to write "a or b or c or d" to sucess).
as you say yourself in the other email, we spend hours every year
discussing these things. solving it sometimes takes less time than
saying we gain nothing (except time) by solving it :)

> 1. QA tools:

witch one doesn't support both existing schemas ?

> 2. Cartography

witch one doesn't support both existing schemas ?

> 3. Videos/blogposts/training material/books

Indeed, a book about osm written in 2011 is partially
obsolete in 2021.
If someone really still use it, it would simply have
a notice "deprecated tag" + one click to learn more about it,
clic to fix. very nice !
it would be wonderful if all colleagues could have such
an easy continuous training.

please discuss the arguments, not your opinion
on the intelligence of those who disagree with you,
nor theoretical cases without giving at least a real case.
fragmentation is an issue, a major one because everybody
need to take care of both (or more) schemas.

Regards,
Marc