Issues with code folding not following expected Rules

I am having an issue where the code folding doesn't seem to be following the normal protocols I expected it to. I expect the code folding to condense functions, for/while loops, and multi-lined comments. But I am seeing cases where the folding is happening from seemingly arbitrary lines.

For reference, I believe this code was originally written earlier this year on MATLAB R2021b but I have upgraded to R2023a in the past few weeks. I didn't notice the code folding issue until just now but I am unsure how long it's been present.

I apologize in advance for the large sections of code but from what I have seen elsewhere, code folding issues are often hard to replicate so I am keeping my original syntax in the code snippets.

Example 1:

Original Code: Lines 258 - 284

% Copy for running mat save file
% AutoSaveResultsByRun(matname,tInfo,vInfo,vals,runs,RunLogs,"OnlyUnchecked",true)


%% Find additional offset to get Kulites to rise at t=0
% This is important so that eventually the spectra calculations can be
% performed on a trace that is as 50/50 balanced as possible. For now,
% assume that the PCB references are aligned enough with the Kulites and
% that the relative delay is constant between the two sets. Therefore, we
% only need to institute the delay discovered by the reference Kulite
[~,itemp1] = min(abs(tshift(:,:,2)+0.1));
[~,itemp2] = min(abs(tshift(:,:,2)-0.1));
% Loop through each Reference Trace to detect the initial shock
iKulRef0 = zeros(R,1); % Indices needed to shift Kulite Data
for i = 1:R
    [iKulRef0(i),res] = findchangepts(dat.Ch2(itemp1(i):itemp2(i),i),...
        'Statistic','std');
end

% Generate the new time trace for the Kulite
tKul0 = tshift(:,:,1:2); % This only needs to have vectors for Ch1 and Ch2
% Find the time shift needed for each run. This should be highly constant
for i = 1:R
    tKul0(:,i,:) = tKul0(:,i,:)-tshift(itemp1(i)+iKulRef0(i)-1,i,2);
end
% Save the full indices of these tKul0 locations
iKulRef0 = itemp1.' + iKulRef0 - 1;

Folded Code: Folds such that Lines 259 and 279 are adjacent and I have the following code visible

% Copy for running mat save file
% AutoSaveResultsByRun(matname,tInfo,vInfo,vals,runs,RunLogs,"OnlyUnchecked",true) 
% Find the time shift needed for each run. This should be highly constant
for i = 1:R
    tKul0(:,i,:) = tKul0(:,i,:)-tshift(itemp1(i)+iKulRef0(i)-1,i,2);
end
% Save the full indices of these tKul0 locations
iKulRef0 = itemp1.' + iKulRef0 - 1;

This code folding goes across a section break and includes both comments and code which makes no sense to me. Also this code folding option doesn't dissappear when I uncomment line 258 which invokes a separate function (in an external script) called AutoSaveResultsByRun. There don't appear to be any code folding issues in that separate function either.

Example 2:

Original Code: within local function part of same script - Lines 460 to 520. The confusion gets way worse as the folding and unfolding in this section is direction dependent because of internal code folds somehow.

function ...
    ...
    ...
% Populate the Channel Vectors --------------------------------------------
for i = R:-1:1 % Loop through each run (Backwards to avoid preallocation)
    for j = 1:4 % Loop through each Channel
        % Based on the Channel, Set reference data
        if j <= 2;  temp = kPa.(['Ch',num2str(j)]); % Kul Ref Matrices
        else;       temp = dat.(['Ch',num2str(j)]); % PCB Ref Matrices
        end
        % Now extract the data and insert into the appropriate substruct
        y = temp(ia(j,i):ib(j,i),i);
        d4spec.(['Ch',num2str(j)])(:,i) = y;
    end
end
% Print that this has been complete
disp('Truncated Data-for-Spectra Calcs has been created')
end % End of function 1

% ======== Find Expansion Wave arrival ====================================
% This should be the end of the upper-step when the expansion waves impinge
% on the sensor array. At this time, a linear decay should begin
% -------------------------------------------------------------------------
% INPUTS:
%       ia: The time index for the lower bounds of the search area
%       ib: the time index for the upper bounds of the search area
%       d: Data structure that has the time and Channels (d4spec)
%       plotExpWaveID: Decide whether to plot these or not
% OUTPUTS:
%       event: Index representing the expansion wave impingement (both Kul)
% -------------------------------------------------------------------------
function event = findExpWaveArrival(ia,ib,d,plotExpWaveID)
global Cruns; global SavePNG; % Call Global values
R = size(d.t,2);    % Find the number of runs
event = zeros(R,2); % Preallocate the event matrix
% First find the event in channel 2 (Reference Kulite) --------------------
for i = 1:R
    % Save temporary vector for now
    temp = d.Ch2(ia(i):ib(i),i);
    % Find the actual indices of the change point, use linear change
    [ipt,~] = findchangepts(temp,'Statistic','linear');
    event(i,2) = ia(i)+ipt-1; % Save that index value
    % Create a temporary legend
    templeg = {sprintf('Run %d: Ref Kul',Cruns.Run_num(i)),...
        'Linear Fits',sprintf('t = %.3f ms',d.t(event(i,2),i))};
    % Generate temporary FileName
    figname = sprintf('Run%d-Ch2-Detect-ExpWaveArrival',Cruns.Run_num(i));
    tempdir = fullfile('Figures','EventDetection',...
        'ExpansionWaveArrival','RefKul');
    figname = fullfile(tempdir,figname);
    % Generate a plot too if desired
    if plotExpWaveID
        % Plot for visual confirmation
        figure('Position',[124,287,764,360]);
        findchangepts(temp,'Statistic','linear')
        legend(sprintf('Run $d',Cruns.Run_num(i))); %#ok<CTPCT>
        % Add some labels
        ylabel('Pressure (kPa)');
        legend(templeg,'Location','best','FontSize',12);
        xlim([0 length(temp)])
        set(gca,'FontSize',12)
        % Save the figure if desired
        if SavePNG; saveas(gcf,figname,'png');  end
    end
end
    ...
    ...
    ...
end

Folded Code: Somehow this fold goes from an if statement, across a large comment block and into the middle of the next local function. I have no idea why this is happening.

function ...
    ...
    ...
% Populate the Channel Vectors --------------------------------------------
for i = R:-1:1 % Loop through each run (Backwards to avoid preallocation)
    for j = 1:4 % Loop through each Channel
        % Based on the Channel, Set reference data
        if j <= 2;  temp = kPa.(['Ch',num2str(j)]); % Kul Ref Matrices
    tempdir = fullfile('Figures','EventDetection',...
        'ExpansionWaveArrival','RefKul');
    figname = fullfile(tempdir,figname);
    % Generate a plot too if desired
    if plotExpWaveID
        % Plot for visual confirmation
        figure('Position',[124,287,764,360]);
        findchangepts(temp,'Statistic','linear')
        legend(sprintf('Run $d',Cruns.Run_num(i))); %#ok<CTPCT>
        % Add some labels
        ylabel('Pressure (kPa)');
        legend(templeg,'Location','best','FontSize',12);
        xlim([0 length(temp)])
        set(gca,'FontSize',12)
        % Save the figure if desired
        if SavePNG; saveas(gcf,figname,'png');  end
    end
end
    ...
    ...
    ...
end

As of now, I haven't seen this behavior in any of my other scripts, but I haven't scoured through them all at this point. Also I hope the lack of context on the code's purpose doesn't hinder the clarity of my question.

Answer 1

[Emma Farnan]

I actually just solved the issue by copy+pasting the entire code into a new script. I changed nothing, but the folding issues were no longer present.

Nonetheless, I am going to leave this question as "unanswered" for the moment so that it can be properly flagged as a potential issue when upgrading to R2023a - where I pressume the issue originated from.

[João GONÇALVES de Oliveira chueire]

Had the same problem and copy+pasting also worked for me

Copied into another file and then back in the original one and it worked fine.

Answer 2

[alexander Mcghee]

So you want MATLAB to be a great IDE, so do I...

try visual studio code

Integrate matlab with

https://github.com/apommel/vscode-matlab-interactive-terminal

and you can use co-pilot from github to help you code more effeciently. Its worth the rabbit hole