Compared to 0.9, 1.0.1 does not handle re-indentation correctly

[PetrKryslUCSD]

For instance, this code is not re-indented correctly when I do "select all" and "reindent":

function e1_hole_q4()

    function solve(iteration, fens, fes)
     
        # Identify the "growth" layer of elements
        node_is_act_bdry = fill(false, count(fens))
        node_is_act_bdry[lopeningn] .= true
        el_in_growth_layer = fill(false, count(fes))
        for i in 1:count(fes)
            for j in fes.conn[i]
                if node_is_act_bdry[j]
                    el_in_growth_layer[i] = true
                    break
                end
            end
        end
        
# Apply mechanical support conditions to eliminate rigid body
# displacements
u = NodalField(zeros(size(fens.xyz, 1), 2)) # displacement field
setebc!(u, lleft0n, 1, 0.0)
setebc!(u, lleft0n, 2, 0.0)
setebc!(u, lright0n, 2, 0.0)
applyebc!(u)
numberdofs!(u)
end
end

ST3 with Julia-sublime package 0.9 can do it right.

[PetrKryslUCSD]

I checked the Indentation.tmPreferences files for ST versions 3 and 4: they are identical. So at the moment I have no explanation for the different behavior. I am in a conversation with the developers of ST to see if we can sort it out.

[PetrKryslUCSD]

I was able to narrow it down quite a bit:

function f(x)
y = x^2 # comment
end

function f(x)
y = x^2 
end

In ST 4 the first definition of the function is not reindented properly. ST 3 does the job correctly.

[PetrKryslUCSD]

It seems to me that being confused about scope could be an explanation.
Or not (consultation with @OdatNurd)....

[PetrKryslUCSD]

More data: sublimehq/Packages#2550 (comment)

[PetrKryslUCSD]

The problem is apparently in the definition of comments:

comments:
    - match: \#=
      scope: punctuation.definition.comment.number-equal-sign.julia
      push:
        - meta_scope: comment.block.number-equal-sign.julia
        - match: =\#
          scope: punctuation.definition.comment.number-equal-sign.julia
          pop: true
    - match: \#+
      scope: punctuation.definition.comment.julia
      push:
        - meta_scope: comment.line.number-sign.julia
        - match: \n
          pop: true

I replaced this with the old definition

  comments:
    - match: '#='
      push: comment-block
    - match: '#.*'
      scope: comment.line.number-sign.julia

  comment-block:
    - meta_scope: comment.block.number-sign-equals.julia
    - match: '#='
      push: comment-block
    - match: '=#'
      pop: true

and the re-indentation on ST 4 works as well as on ST3.

Edit: Actually, with the new definition re-indentation doesn't work when a block comment is present. Neither on ST 4, nor on ST3.
And, being the "old" definition, it never worked with ST3 for block comments.

[PetrKryslUCSD]

It seems that @ViktorQvarfordt is the author of the comment definitions? Any ideas how to fix this properly, Viktor?

[PetrKryslUCSD]

Except for the issue of the block comments the proposed change makes the indentation work the same in ST 4 and ST 3.

[randy3k]

I always do no trust the auto-indentation feature of editors. If you want to format a document, it is better to use a code formatter, for example https://github.com/domluna/JuliaFormatter.jl

[PetrKryslUCSD]

That is true, I tried that package and it works great. However, the workflow is much smoother with re-indentation integrated into the editor. And most of the times it works just fine.

[PetrKryslUCSD]

By the way, I use re-indentation most often on just selections, not the entire documents. Then the use of a package to do that is much less attractive.

[randy3k]

I have pushed one more commit, all the comments code are ignored when applying indentation.

function f(x)
# comment
x = 1
end

function(x)
        #comment
        x = 1
end

function(x)
#=
comment 1
    comment 2
=#
x = 1
end

function(x)
        #=
        comment 1
            comment 2
        =#
        x = 1
end

becomes

function f(x)
# comment
    x = 1
end

function(x)
        #comment
    x = 1
end

function(x)
#=
comment 1
    comment 2
=#
    x = 1
end

function(x)
        #=
        comment 1
            comment 2
        =#
    x = 1
end

[PetrKryslUCSD]

I don't know, this is what I got:

function f(x)
# comment
x = 1
end

function(x)
        #comment
x = 1
end

function(x)
#=
comment 1
    comment 2
=#
x = 1
end

function(x)
        #=
        comment 1
            comment 2
        =#
x = 1
end

Edit 10/15: I don't know what happened last time, but when I redid this whole process of installing the PR, I got this result:

function f(x)
    # comment
    x = 1
end

function(x)
    #comment
    x = 1
end

function(x)
#=
comment 1
    comment 2
    =#
    x = 1
end

function(x)
        #=
        comment 1
            comment 2
            =#
            x = 1
        end

[randy3k]

Did you fetch the PR?

[PetrKryslUCSD]

I downloaded the PR and made a package file from it. It did not work for me.

Here's another example where the re-indentation does not work yet:

function flint_loadmesh(meshfile, nunref = 0, nref = 0)
  # Mesh file in the format of the Abaqus .inp file
  fens, fes = BEMScat.MeshingUtilities.loadmesh(meshfile)
  
  if nodesperelem(fes) == 10
    fens, fes = T10toT4(fens, fes) # Convert to linear tetrahedra
end

# The coordinates are in millimeters
fens.xyz .*= phun("mm") # The input file lists the coordinates of the nodes in millimeters
end

[randy3k]

It gives me

function flint_loadmesh(meshfile, nunref = 0, nref = 0)
  # Mesh file in the format of the Abaqus .inp file
  fens, fes = BEMScat.MeshingUtilities.loadmesh(meshfile)
  
  if nodesperelem(fes) == 10
    fens, fes = T10toT4(fens, fes) # Convert to linear tetrahedra
  end

# The coordinates are in millimeters
  fens.xyz .*= phun("mm") # The input file lists the coordinates of the nodes in millimeters
end

which is what I expected.

[PetrKryslUCSD]

Okay, then I am perplexed.
However, it does work when I decrease indents until everything is at the leftmost column.
Then I get

function flint_loadmesh(meshfile, nunref = 0, nref = 0)
# Mesh file in the format of the Abaqus .inp file
    fens, fes = BEMScat.MeshingUtilities.loadmesh(meshfile)

    if nodesperelem(fes) == 10
        fens, fes = T10toT4(fens, fes) # Convert to linear tetrahedra
    end

# The coordinates are in millimeters
    fens.xyz .*= phun("mm") # The input file lists the coordinates of the nodes in millimeters
end

[PetrKryslUCSD]

When I simply copy the code

function flint_loadmesh(meshfile, nunref = 0, nref = 0)
  # Mesh file in the format of the Abaqus .inp file
  fens, fes = BEMScat.MeshingUtilities.loadmesh(meshfile)
  
  if nodesperelem(fes) == 10
    fens, fes = T10toT4(fens, fes) # Convert to linear tetrahedra
end

# The coordinates are in millimeters
fens.xyz .*= phun("mm") # The input file lists the coordinates of the nodes in millimeters
end

and try to re-indent it, then nothing happens.

[randy3k]

I wonder if you have applied the PR correctly.

[PetrKryslUCSD]

I downloaded the PR using Github desktop.

[PetrKryslUCSD]

[PetrKryslUCSD]

I checked the TMPreferences files as well. I do seem to have all your changes.

[PetrKryslUCSD]

Would you like to try it out with my portable ST 4?
https://www.dropbox.com/s/c5gfx2z9xiuusem/archived-SublimeText4_portable-10-15-2020-02-42.zip?dl=0

[PetrKryslUCSD]

function flint_loadmesh(meshfile, nunref = 0, nref = 0)
  # Mesh file in the format of the Abaqus .inp file
  fens, fes = BEMScat.MeshingUtilities.loadmesh(meshfile)
  
  if nodesperelem(fes) == 10
    fens, fes = T10toT4(fens, fes) # Convert to linear tetrahedra
end

# The coordinates are in millimeters
fens.xyz .*= phun("mm") # The input file lists the coordinates of the nodes in millimeters
end

Sorry to report but this still does not work correctly. When I try to re-indent, nothing happens. It stays the same. The other examples above work fine.

[PetrKryslUCSD]

#90 (comment): this is still not reindented properly.

[PetrKryslUCSD]

@randy3k : If the above sample re-indents properly in your ST, perhaps there's something different between our setups? Perhaps another package interferes?