110903 - new script hook not called for fastloaded scripts

Status: VERIFIED FIXED

(Core :: JavaScript Engine, defect)

Description

[Robert Ginda]

patch by brendan coming up.  I've tested debug/linux and noticed no new regressions.

Brendan, I've never seen the xdr code before, could you explain what's going on
with the script magic in the patch?

Comment 1

[Robert Ginda]

reassigning to myself.

Comment 2

[Robert Ginda]

Attachment: patch

Comment 3

[Robert Ginda]

Ignore the makefile change, that was my sloppy diff.

Comment 4

[Brendan Eich [:brendan]]

I simply made the hasMagic out parameter optional: when null, js_XDRScript
reports magic-number-mismatch errors and fails; when non-null, it returns false
if the magic was not a known cookie via *hasMagic, true otherwise in *hasMagic
-- and succeeds (returns true).

I *thought* I wrote a comment in jsscript.h just before js_XDRScript's prototype
(yes, there it is! :-).

Rob, how about r=?  Shaver'll sr= I bet.

/be

Comment 5

[Brendan Eich [:brendan]]

Forgot to say that each serialized script begins with a four-byte magic number,
or cookie, to help catch decoder vs. encoder version/bug mismatch (where either
the decoder is downrev and can't handle a new magic; or the two simply do not
agree on the order and content of bytes in a script's serialization, due to a
bug in their impls, or even in some surrounding serialization code that embeds a
JS script in a larger stream).

/be

Comment 6

[Mike Shaver (:shaver emeritus)]

Comment on attachment 58471 [details] [diff] [review]
patch

sr=shaver.  Nice comment. =)

Comment 7

[Robert Ginda]

Comment on attachment 58471 [details] [diff] [review]
patch

r=rginda, I'll check it into the trunk.

Comment 8

[Robert Ginda]

checked into trunk

Comment 9

[Phil Schwartau]

Verified Fixed -