115695 - Pretty Print Support

Status: RESOLVED FIXED

(Other Applications Graveyard :: Venkman JS Debugger, defect, P1)

Description

[Robert Ginda]

A month or so ago I started working on a "quick hack" to get pretty print
working.  I've completed the native work, but landing the UI would break a few
assumptions made in the front end.

It doesn't make sense to burn this much time on a 'neat' feature when there are
other higher priority bugs to fix, so I'd like to land the native work (so I
don't lose it, and it doesn't rot) in a way that doesn't affect the front end.

Patch coming up shortly.

Comment 1

[Robert Ginda]

Attachment: native changes

There are a few things going on in this patch...
Leak fixes:
  * Script objects were being leaked from enumerateScripts.  Switched to a
COMPtr.
  * Script objects could also be leaked if the debugger were shut down while
there |DeadScript| records waiting to be processed.
  * Script objects could be leaked if there was no scriptHook present when the
script when the script was deleted.

Pretty Print Support:
  * jsdIPC is gone.  All PCs are unsigned long offset from the first PC in the
script.
  * Added JSD_GetJSScript and JSD_GetJSFunction functions.
  * added |pcmap| parameter to pcToLine, lineToPc, and isLineExecutable methods
on jsdIScript.	Selects whether to use the scource code PC map, or the pretty
print map.
  * Added a functionSource property to jsdIScript objects.

Added |tag| property to jsdIScripts.  This gives me something to use as a hash
key in JS in the future.

A UI patch will appear shortly to take care of the affected callsites.

Comment 2

[Robert Ginda]

Attachment: ui changes

Adds View->Pretty Print menu item.  Breakpoints cannot be set from pretty print
mode.

Comment 3

[Brendan Eich [:brendan]]

rs=[e-mail] on both patches, they look good at a glance (wow, uint64
pc offsets?  Never mind the debugger, SpiderMonkey isn't ready for such
gigundo-scripts).

/be

Comment 4

[Robert Ginda]

Checked in.