Doc links no longer work

I wrote a toolbox sometime ago. It had voluminous documentation. This used to work: doc AETNA. It would bring up a window with the documentation, and clicking on provided links would lead to further information.

This is now broken. How do I get back the previous behavior?

[Steven Lord]

What does "broken" mean in this context?

• Do you receive warning and/or error messages? If so the full and exact text of those messages (all the text displayed in orange and/or red in the Command Window) may be useful in determining what's going on and how to avoid the warning and/or error.
• Does it do something different than what you expected? If so, what did it do and what did you expect it to do?
• Did MATLAB crash? If so please send the crash log file (with a description of what you were running or doing in MATLAB when the crash occured) to Technical Support so we can investigate.

[Image Analyst]

Is your preferences for Help set up to look for help on the internet (the default) or locally? If you install the local help documentation, and then try your toolbox help, does it work then?

[Petr Krysl]

Thank you for your response.

Help/documentation for my toolkit used to show up in a window like this one in response to the command doc AETNA:

Then I could click on the individual links to lead me to further pages.

Now in order to get to this page, I had to do "help AETNA" which dumped this page in the command line. And then I had to control-click on one of the links to get to this window. But none of the links work.

Doing "doc AETNA" leads to a webpage

which is obviously not going to do any good, since my help files are not online.

So, since this worked with R2021a (and I believe, but I'm not sure, R2023a), you guys did something to break these help files.

How does one fix this?

Thanks.

[Walter Roberson]

Right at the moment there is a back-end failure on the online documentation server, which is being repaired.

I don't think you were affected, however: I think you posted this question before the back-end failure started.

[Walter Roberson]

The back-end server is back in action. You could try again. (I don't think it will make any difference for your purpose, though.)

[Petr Krysl]

The help system of the toolkit is based on Contents.m files and the source directories provide the structure. I'm not aware of this having been made obsolete?

[Petr Krysl]

When I use Matlab in the cloud, and I do 'doc AETNA', I get this web page.

Which is hilarious and sad at the same time.

[Petr Krysl]

Just checked with R2021a: everything works, all the links in the documentation open new tabs in the help browser.

R2024a: doc AETNA opened the main browser window, but none of the other links work (no further tabs are opened).

R2024b: nothing works, no help browser is opened.

That seems like a pretty crappy quality control.

[Image Analyst]

Why do you think MATLAB in the cloud would have any idea at all about a toolbox you wrote sometime ago? I know it would not have any idea at all about the code I wrote on my computer.

[Petr Krysl]

I'm afraid you misunderstand: I am saying that Matlab should not have sought the documentation in the cloud, but rather should have served the help documentation stored in my Contents.m files.

[Paul]

@Petr Krysl

I've never used this Contents.m feature before, but I just tried a quick trial in 2024a Update 1 and didn't experience any problems. Didn't really try to push it too hard.

Did you try the suggestion by @Image Analyst to see if choosing the locally installed documentation in Preferences helps?

The doc browser is no longer provided as of 2024b. I wonder if that's a factor in the issues with 2024b.

[Petr Krysl]

Yes, I did play with the settings. No effect.

> The doc browser is no longer provided as of 2024b. I wonder if that's a factor in the issues with 2024b.

Now, this is interesting. The documentation seems not to mention this: in fact, it still reads as if Contents.m was fully supported.

I am at this point really fed up with these incompatibilities.

[Paul]

Yes, the doc page in 2024b does imply that Contents.m is still supported. The 2024b release notes talk about the removal of the doc browser, but makes no mention of any implications on Contents.m functionality. Of course, even if that's an issue for 2024b it doesn't explain why you're also having problems with 2024a.

When I tried doc myfolder in 2024a, the Contents opened in a window like you showed above, and when I clicked a link for a file, that same window transitioned to the help for that file, i.e., not tabbed browsing. I only mention this because you stated above that in 2024a "(no further tabs are opened)" in case there's a difference between what I'm seeing now and what you used to see in prior versions.

[Petr Krysl]

This is typical of the way Mathworks treats the programmers that create its ecosystem: high-handed disregard.

I am done with Matlab. I will rewrite my only package that I still use in the classroom in Python and good riddance Mathworks.

Thank you all for your help.

Answer 1

[Esther]

Doc on a folder with a Contents.m file is not currently working correctly. As a workaround, the undocumented function helpwin can be used instead of doc for these folders till we get this bug fixed.

Answer 2

[埃博拉酱]

This should be a bug. According to Contents.m's documentation, it should work well with the doc command.

If this feature is very important to you, I recommend that you contact technical support. Or you can step through the doc command in both the old and new versions of MATLAB to observe the differences, and then modify the new version of the code.

Update 20241118:

My technical suppoert replied:

This is a regression in the doc command in R2024b. Doc on a folder with a Contents.m file is not currently working correctly.

As a workaround, the undocumented function "helpwin" can be used instead of doc for these folders.

Actually helpwin IS documented, but just deprecated. It works fine in my environment.

[Petr Krysl]

I think I just did contact someone who can do something about the problem (@Steven Lord above)? If that is indeed a bug, and not another incompability foisted on unsuspecting Matlab programmers...

[Steven Lord]

Please contact Technical Support directly using this link. They can work with you and/or the developers responsible for the help and documentation systems to determine the root cause and see if there are any workarounds for this problem. [I don't work on the help and documentation systems except as a consumer of the help and doc.]

[Petr Krysl]

Thank you for the suggestion. I did open a ticket. Their suggestion was to use R2024a. Alas, that one is broken too. How exactly they test the releases is a puzzle to me.

Answer 3

[Matt J]

I experienced this and found it was because of name conflict between one of my user-authored mfiles and a MathWorks-authored mfile that had been introduced in a newer Matlab version. Either matlabrc.m or some other startup Mcode made use of the command and, because it was using my version instead of the MathWorks-authored version, failed to set up properly.

The solution to getting doc running properly would be to make sure all of your user-authored mfiles are below MathWorks-authored mfiles in the path. However, this means your version of the mfile command will be shadowed. If your personal code depends on this command, you could probably track it down by running it in the debugger and waiting for a crash.

[Petr Krysl]

Which file was it?

[Matt J]

For me it was a command called strip() but that doesn't mean your case is being caused by the same file.

[Petr Krysl]

Thanks. I don't think conflicts are my problem.

[Paul]

@Matt J

How did you determine the issue was a name conflict, and how did you determine that strip was the offender? Might be useful information for others who might be having similar/same problem, even if it's not the OP's problem.

[Matt J]

@Paul It's too long ago. I wouldn't be able to reconstruct the troubleshooting steps. I imagine if you create a startup.m file and insert,

dbstop if caught error

and then relaunch Matlab, it might be a way to trap Matlab startup problems, even if that wasn't precisely what I did.