deprecate syntax 1.+. fixes #19089

[JeffBezanson]

No description provided.

[Sacha0]

lgtm (modulo the parser changes which I did not pore over)! :)

[Sacha0]

Might be worth mentioning the need for disambiguation in the documentation somewhere?

[JeffBezanson]

Added a bit on this to the manual.

[Sacha0]

lgtm for the manual addition! :)

[jebej]

Was there any argument for why keeping the 1. syntax is preferable? The only two I see in #19089 was that other languages accept it, and that it would be annoying to have people replace all the dots with .0. This does not seem super convincing, and people in that thread also seemed to prefer to disallow 1..

I find that putting spaces for operations like division, multiplication and exponentiation to be fairly jarring as it makes their higher precedence less obvious.

I also think more people use the syntax 1.*[1,2,3]./6 .+ 5 than use the 1. syntax, ie more people will be affected by this change than by disallowing 1.. This is evidenced by this blurb in the manual:

 julia> 1 + 2 + 3
6

julia> 1 - 2
-1

julia> 3*2/12
0.5

(By convention, we tend to space operators more tightly if they get applied before other nearby operators. For instance, we would generally write -x + 2 to reflect that first x gets negated, and then 2 is added to that result.)

[JeffBezanson]

I see numeric literal syntax as much more established than dot operator syntax. Even if 1. were disallowed in julia, it would be easy to read 1.+x as (1.)+x due to the prevalence of 1. as numeric literal syntax. In any case, deprecating 1.+x is the safest thing, since that at least allows us to change its meaning in the future.