On 2014/08/29 00:34:18, gholap wrote:
One issue here is:
Most of these code insertions like
isolate->CodeCreateEvent(...)
can be avoided if we use JitCodeEventHandler instead of writing our own
event
handler.
What stops us from doing that?
The function names JitCodeEvent are not exactly same as those that are
printed
in v8.log
For example:
code-creation ... 0x24b36585e8a0,268,LazyCompile:~funcname :1
(JitCodeEvent)
code-creation ... 0x24b36585e840,364,funcname :1:99 (v8.log)
The handler in this CL tries to follow v8.log as faithfully as possible.
Also should be noticed that the starting addresses as well as the size of
the
compiled code is reported differeted by the two loggers.
I would strongly oppose introducing a whole new API if we already almost
have
what we need. What needs to be investigated:
* What are the detailed differences between your proposed API and what we
already have in v8?
* Are the differences *really* important or could those be handled on the
embedder side?
* If they are important, that's not a good reason to come up with a
whole new
duplicated API. We should tweak what we have already to cover the new use
case.
https://codereview.chromium.org/519543002/
--
--
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
---
You received this message because you are subscribed to the Google Groups "v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
For more options, visit https://groups.google.com/d/optout.
