[Webkit-unassigned] [Bug 64468] WIN: DumpRenderTree hangs under NRWT

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=64468





--- Comment #21 from Dirk Pranke <dpranke at chromium.org>  2012-01-05 12:34:01 PST ---
(In reply to comment #18)
> (In reply to comment #14)
> > (In reply to comment #13)
> > > Your proposed change in comment 7 isn't quite right: tests in http/tests/local are supposed to be loaded as local files, not as http: URLs. At least according to Port.test_to_uri.
> > >
> > 
> > Oh. Hm. You're right. That's bad, as it means the logic in controllers/manager.py is broken, and we should probably create an is_http_test() to actually codify/fix this.
> 
> Sounds good to me! I think http/tests/websocket/tests/local is supposed to be run locally too.
>

Actually, now that I think about it further, the manager logic is correct; it needs to know whether a server is needed, not whether the test itself is local or http.

> Does Driver know whether it's running under Cygwin or not? I didn't think it did.
>

It certainly could know if it needed it ...

> > By that same logic, test_to_uri() is only used by the chromium Driver (and various pieces of test code), and should probably be moved out of the Port object and down into the chromium Driver.
> 
> It's also used by WebKitPort._command_from_driver_input, of course!

right :) Even so, this should probably be on the Driver class and not the port class if nothing outside of Drivers need it.

(In reply to comment #19)
> We could probably change things such that Apache is always running whenever any test is run. Then these tests could be moved out of the http/ directory entirely.

In addition to Eric's comment, it also makes NRWT's startup significantly slower in the common case of just running a subset of non-http tests.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.