Show frame skip status in call stack

[roblourens]

If a frame is for a skipped file, show (skipped) in its name.

[roblourens]

@isidorn and @weinand, I'm thinking about how to show that a file is skipped in the call stack. Useful for a few reasons,

• The feature is sort of invisible otherwise
• Something visible will happen when you use the context menu option (I can resend the pause event, so the call stack gets refreshed)
• The context menu must be a toggle because of how the API works, so you need to know the initial state

I can add a tag to the left or right side:

The right side might be better since it's the file that's skipped, not the function. But if the frame's Source has a path, then it displays part of the path instead of the source name, so I can't customize it. I guess this is related to #16913. Would it be a breaking change to show the name on the right side, if it's set? Other ideas?

[isidorn]

I would start with showing the (skipped) on the right hand side. It belonds next to the file name since the file is skippped and (skipped) is better in a regular than verbatim font.
If that does not work then we can tune it

[roblourens]

Sure, so does it make sense to always show the source name when available? Should I open an issue for that?

[weinand]

Instead of showing 'skipped' we could try a font variation that does not clutter the UI that much, e.g. showing the module name dimmed and the hover would explain the reason for this).

[roblourens]

Then the protocol would have to know about skipFiles?

[weinand]

So your approach with the "(skipped)" prefix would just prefix the name attribute of the Source object?

[roblourens]

Yeah (I would really like the font variation though, it's more similar to what other tools do)

[roblourens]

I was debugging some async/await code and was thinking about how much nicer this would look if the module name was dimmed on non-sourcemapped frames

Those frames are skipped with smartStep, not skipFiles, but the two should be marked the same in UI. And was also thinking that I could change the smartStep implementation to be based on the blackbox API, for performance.

[weinand]

Even if we are not adding the 'skipFiles' concept to the debug protocol, we could add some abstract attribute to the StackFrame or Source type.
@roblourens any suggestion how we could name the attribute?

[roblourens]

hidden or dimmed? Hm, can you think of other situations where this would be useful?

[weinand]

@roblourens 'hidden' sounds good. 'dimmed' is too concrete because it is a real visual style.
C++/C# could use 'hidden' to indicate that source is not available.

[roblourens]

I'm trying to think of a good synonym for deemphasized since it's technically more accurate than dimmed.

[weinand]

@roblourens I like 'deemphasize' (and the word seems to exist: http://www.thesaurus.com/browse/de-emphasize but I didn't like any of the listed synonyms).
I've created a protocol change request for this: microsoft/vscode-debugadapter-node#93

[weinand]

@roblourens I've added a presentationHint attribute to Source and a value deemphasize. It is available in version 1.16.0-pre.1 of the debug protocol.

[roblourens]

This looks great! Thanks guys.

Now I will use the same UI for code that was smartStep'd over. I'll use the tag (smartStep) instead of (skipped).

[roblourens]

Yes, a stack from async/await with smartStep is infinitely more readable now: