822563 - JS GDB pretty-printers: pretty-print JSObject & values

Status: RESOLVED FIXED

(Core :: JavaScript Engine, defect)

Description

[Jim Blandy :jimb]

Attachment: Pretty-print references to JSObject and its subclasses.

The GDB pretty-printer should pretty-print references to JSObject instances (including instances of classes derived from JSObject).

That is, instead of this:
$16 = (js::ScopeObject &) @0x7ffff152fc40: {<JSObject> = {<js::ObjectImpl> = {<js::gc::Cell> = {<No data fields>}, shape_ = 0x7ffff15210d8, type_ = 0x7ffff151f040, slots = 0x0, elements = 0xc00510}, }, }

we should get this:
$16 = (js::ScopeObject &) @0x7ffff152fc40: [object Call]

Comment 1

[Jim Blandy :jimb]

Attachment: Pretty-print references to JSObject and its subclasses.

New patch; original included stray changes from other work.

Comment 2

[Jim Blandy :jimb]

Attachment: Pretty-print references to JSObject and its subclasses.

One more try. I was careless in the pretty-printer lookup cleanup. Includes a regression test.

Comment 3

[Steve Fink [:sfink] [:s:]]

Comment on attachment 693266 [details] [diff] [review]
Pretty-print references to JSObject and its subclasses.

Review of attachment 693266 [details] [diff] [review]:
-----------------------------------------------------------------

This worked out pretty nicely, unless there's something horrible I'm missing when eyeballing this.

::: js/src/gdb/mozilla/prettyprinters.py
@@ +312,4 @@
>  class Pointer(object):
>      def __new__(cls, value, cache):
>          # Don't try to provide pretty-printers for NULL pointers.
> +        if value.type.code == gdb.TYPE_CODE_PTR and value == 0: return None

nit: I like the same-line return for very short conditionals, but I find this one hard to spot.

Comment 4

[Jim Blandy :jimb]

Thanks for the review! I changed the same-line return.

https://hg.mozilla.org/integration/mozilla-inbound/rev/6b17e37a0762

Comment 5

[Daniel Holbert [:dholbert]]

Backed out for build error on Windows:
 https://hg.mozilla.org/integration/mozilla-inbound/rev/f29149459b0f

The error looks like this:
{
e:/builds/moz2_slave/m-in-w32/build/js/src/gdb/tests/test-JSObject.cpp(23) : error C2027: use of undefined type 'JSObject'
}
https://tbpl.mozilla.org/php/getParsedLog.php?id=18139695&tree=Mozilla-Inbound
https://tbpl.mozilla.org/php/getParsedLog.php?id=18140081&tree=Mozilla-Inbound

Comment 6

[Daniel Holbert [:dholbert]]

(Clearing target milestone & in-testsuite for the moment, due to backout.)
(Incidentally: you don't have to bother setting Target Milestone when pushing to mozilla-inbound -- our merge-inbound-to-central-and-comment-on-all-the-bugs tool will automatically will do that for you, I'm told.)

Comment 7

[Jim Blandy :jimb]

"Surely this won't need a try push."

Comment 8

[Jim Blandy :jimb]

Try: https://tbpl.mozilla.org/?tree=Try&rev=12ce7e3c1409

Comment 9

[Jim Blandy :jimb]

Okay, that try run looks clean. Landing again:
https://hg.mozilla.org/integration/mozilla-inbound/rev/723e403c36b1

Comment 10

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

https://hg.mozilla.org/mozilla-central/rev/723e403c36b1