string cell array inside a graph

Say I have this:

C = {'a';'b';'c';'d';'e'}; 

s = [1 1 2 3 1];
t = [2 2 3 4 4];
G = graph(s,t);
G.Edges.Label = C;

With the following two commands I would expect to get the same output, but I get different, why with the second comand only one output is given?

C{[2 3]}
ans =

    'b'


ans =

    'c'

G.Edges.Label{[2 3]}

ans =

    'b'

[Stephen23]

Similar to:

https://www.mathworks.com/matlabcentral/answers/344486-generate-comma-separated-list-in-single-line-of-code

[Rub Ron]

with subref is clear because it is a function. I would not think of G.Edges.Label as a function till now. Looks strange.

Answer 1

[Bruno Luong]

I'm surprise no one give you get an answer. There is some subttle internal working needed to be explained.

The syntax

C{[2 3]}

returns a comma list. No function is called.

The second command

G.Edges.Label{[2 3]}

call a graph method (probably an overloaded subsref) and it returns on the LHS multiple output IF LHS it's provided. If not the default LHS and Matlab ANS (single "variable").

However if users provide 2 LHS they'll get them.

>> [bb,cc] = G.Edges.Label{[2 3]}

bb =

    'b'


cc =

    'c'

>> 

This behavior is very similar to any multiple-output function such as

>> [a,b]=meshgrid(1:3,1:2) % Get 2 LHSs

a =

     1     2     3
     1     2     3


b =

     1     1     1
     2     2     2

>> meshgrid(1:3,1:2) % get only the first LHS 

ans =

     1     2     3
     1     2     3

To digest all that think about the following result

 >> G.Edges.Label([2 3]) % parenthesis

ans =

  2×1 cell array

    {'b'}
    {'c'}

>> ans{:}

ans =

    'b'


ans =

    'c'

>> 

[Rub Ron]

it is a surprise that G.Edges.Label in G.Edges.Label{[2 3]} is treated as a function. It loos like a struct variable

[Bruno Luong]

Expression like this is called by the overloaded subsref function of GRAPH classdef.

However more confusing is this, if you do this, it will return properly

>> E=G.Edges

E =

  5×2 table

    EndNodes    Label
    ________    _____

     1    2     {'a'}
     1    2     {'b'}
     1    4     {'c'}
     2    3     {'d'}
     3    4     {'e'}

>> E.Label{[1 3]}

ans =

    'a'


ans =

    'c'

The first command call graph subsref to return a TABLE

Then the second command call a table subrefs and it return correctly a comma list. The question is how table manages to do that is a little mysterious to me. It might have more sophisticated SUBREFS and might be an intermediate true CELL variable is created before the final result.