[stdlib_io] disp(display your data)

[zoziha]

• to_string is merged.
• unit argument: output_unit
• brief argument.
• how to handle the details of brief optional argument for users (see [stdlib_io] disp(display your data) #445 (comment))
• default format (see [stdlib_io] disp(display your data) #445 (comment))
• to support string_type in stdlib_string_type.fypp (see [stdlib_io] disp(display your data) #445 (review))
• If possible, support for linked list(Stdlib linked list #491) in disp would be a good addition. I am following it. (better not)

This branch develops subroutines disp and format_string for stdlib based on the disp and num2str functions in the forlab library.
So, now disp API is:

call [[stdlib_io(module):disp(interface)]]( [value, header, unit, brief] )

disp subroutine

Easily display strings, scalars and rank-1/rank-2 arrays to the screen (or another output unit).

Effects

 disp(string):
 string
 It is a note.

 disp(r):
 [matrix size: 2×3]
 -0.1000E-10  -0.1000E+11    1.000
   1.000        1.000        1.000
 disp(c):
 [matrix size: 2×3]
            (1.000,0.000)             (1.000,0.000)             (1.000,0.000)
            (1.000,0.000) (-0.1000E+11,-0.1000E+11)             (1.000,0.000)
 disp(i):
 [matrix size: 2×3]
           1            1            1
           1            1            1
 disp(l):
 [matrix size: 10×10]
           T            T            T          ...            T
           T            T            T          ...            T
           T            T            T          ...            T
           :            :            :            :            :
           T            T            T          ...            T
 disp(c_3d, 3):
 [matrix size: 2×100]
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
 disp(c_3d, 2):
 [matrix size: 100×20]
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
                        :                         :                         :                         :                         :  
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000) 

format_string function

(see #444)

Links

1. (issue see Function to format variable as string #435 )🔰
2. keurfonluu/Forlab
3. zoziha/forlab
4. Add complex number Descriptor j3-fortran/fortran_proposals#212

[awvwgk]

Thanks @zoziha, @St-Maxwell for opening those pull requests. I think it is fine to have two PRs here, we will first discuss on #444 about the string formatting routines and once we agree on the API we will continue with the review of the display function. How does the plan sound to you?

[zoziha]

Strange discovery

[jvdp1]

Thank you for this PR.
It might be good to submit 2 PRs, instead of one. It could help the reviewers.

[St-Maxwell]

I may have some questions about the scheme of setting output_unit:

1. A single * Already means the default output location, which is the same as [iso_fortran_env(module):output_unit(integer)] and belongs to the standard output, or is it that I understand it wrong, they are different?
2. Or your idea is to add a input argument for setting the output_unit number to the disp subroutine.
   If it is the latter, you will encounter such a problem, and the polymorphic interface with two optional arguments will encounter such a problem (see fortran interface optional). At this time, it is better to consider rewriting a set of functions disp(value [, string]) + disp(value, unit [, string]). (I can't express what I mean very well)

The existing test (comment) does not report an error if there is any issue, because it simply outputs data to the screen (output_unit) and cannot be tested well. If the output_unit scheme is improved, further testing may be possible.

I think the interface of disp would be something like disp(value [, unit=output_unit] [, string]). The dummy arguments unit and string are optional, and unit has a default value output_unit.
In fact, disp reminds me of raw_print function in Armadillo. That's why I proposed a similar interface for disp.

[arjenmarkus]

The parameter output_unit is equivalent to the * in "write(*,'(a)') 'Hello'". Historically the value was 6, so that you may encounter "write(6,*) ..." as well, but there is NO guarantee that LU-number 6 corresponds to the console. Before the introduction of output_unit and its brethren it was not really possible to refer to "standard output" etc in a standard way. Op di 6 jul. 2021 om 05:50 schreef zoziha ***@***.***>:

…

I may have some questions about the scheme of setting output_unit: 1. A single * Already means the default output location, which is the same as [iso_fortran_env(module):output_unit(integer)] and belongs to the standard output, or is it that I understand it wrong, they are different? 2. Or your idea is to *add a input argument* for setting the output_unit number to the disp subroutine. If it is the latter, you will encounter such a problem, and the polymorphic interface with two optional arguments will encounter such a problem (see fortran interface optional <https://stackoverflow.com/questions/3121954/fortran-90-presence-of-optional-arguments>). At this time, it is better to consider rewriting a set of functions disp(value [, string]) + disp(value, unit [, string]). (I can't express what I mean very well) The existing test (comment <#445 (comment)>) does not report an error if there is any issue, because it simply outputs data to the screen (output_unit) and cannot be tested well. If the output_unit scheme is improved, further testing may be possible. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#445 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AAN6YRYCMED5L5Q3RKBBLF3TWJ4ORANCNFSM47KNB3SA> .

[zoziha]

Hello, I got a suggestion from @St-Maxwell and tried to implement it simply. I created a new branch disp_unit_brief in which I implemented optional unit and brief. If all of you feel good, I can merge it to this PR.

Example

(see stdlib_io.md/disp_example)

 disp(string):
 string
 It is a note.

 disp(r):
 [matrix size: 2×3]
 -0.1000E-10  -0.1000E+11    1.000
   1.000        1.000        1.000
 disp(c):
 [matrix size: 2×3]
            (1.000,0.000)             (1.000,0.000)             (1.000,0.000)
            (1.000,0.000) (-0.1000E+11,-0.1000E+11)             (1.000,0.000)
 disp(i):
 [matrix size: 2×3]
           1            1            1
           1            1            1
 disp(l):
 [matrix size: 10×10]
           T            T            T          ...            T
           T            T            T          ...            T
           T            T            T          ...            T
           :            :            :            :            :
           T            T            T          ...            T
 disp(c_3d, 3):
 [matrix size: 2×100]
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
 disp(c_3d, 2):
 [matrix size: 100×20]
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000)  
                        :                         :                         :                         :                         :  
            (2.000,0.000)             (2.000,0.000)             (2.000,0.000)                       ...             (2.000,0.000) 

Reference links

• armadillo/brief_print
  

• ctsa/mdisplay

[zoziha]

I found two problems during my coding process:

1. Other data types, such as integer, complex, character and logical types, are all right-aligned, but the default printing format of real types is neither left-aligned nor right-aligned. This looks weird in disp_brief, of course, we can ignore it, this problem is not fatal.
   
2. I saw stdlib set the derived string type string_type, I wonder if disp needs to support it?

[zoziha]

Add support for string_type type initially, and ready to get your feedback and update docs.
This PR is for stdlib_io_disp routines, #444 PR is for stdlib_string_format_string routines.
Thank you all~

[zoziha]

Because to_string has been merged and disp depends on to_string, I updated disp this time.
Come to think of it, disp seems to have enough possibilities. We welcome your suggestions for extension and modification.
The purpose of disp is to help users easily know the matrix data and print statements.

call disp("It is a note.")
call disp(x, "I am a scalar:")
call disp(A(1:10, 2:4), "Let me see the array:")