One approach we've used in other code is to put the word "deprecated"
in the name of the function in question. That makes these issues very
visible in code reviews. For a constructor, you can require an enum
value with depreciated in its name.
Adam
On Wed, Jan 19, 2011 at 6:11 PM, Ryosuke Niwa
Hi everyone,
Please DO NOT instantiate a Position object by Position::Position(PassRefPtr
anchorNode, int offset) or call Position::node().
The above constructor creates a legacy editing position, which we're trying
to get rid of [1]. Please use Position(PassRefPtr anchorNode, int
offset, Ancho
There was a corrupt file in the webserver cache which has been removed. Please
let me know if anyone hits this again. The problem was not specific to sdch,
but headers are used to hash cache files so only a specific set of headers got
you the bad file.
-Bill
On Jan 19, 2011, at 2:24 PM, Mik
It is not necessary to specify a category.
http://build.webkit.org/console?reload=60 also works (at least, for
me).
On Thu, Jan 20, 2011 at 01:20, James Robinson wrote:
> http://build.webkit.org/console fails to load in Chromium (and Google
> Chrome) today. It seems that the bug is in the server
http://build.webkit.org/console fails to load in Chromium (and Google
Chrome) today. It seems that the bug is in the server's Accept-Encoding:
handling. If the Accept-Encoding line includes 'sdch' then the server
closes the socket without sending any response bytes. Requests for other
pages like
Right ;)
For now, I ended up implementing the functionality differently in a
way that doesn't use the os.walk() routine. This is in some sense
suboptimal, so I'm still curious to know what the answer is.
-- Dirk
On Tue, Jan 18, 2011 at 10:11 PM, Adam Barth wrote:
> In the past, we've used autoi
Those instances are as much likely to be WebCore bugs as DRT bugs.
DRT may not be properly clearing state between tests, but it's equally
likely that WebCore is bleeding state between pages (including doing
things like smashing memory).
On Wed, Jan 19, 2011 at 3:59 AM, Osztrogonac Csaba wrote:
>
On Jan 18, 2011, at 11:28 PM, Maciej Stachowiak wrote:
> Off-topic, but... we really should change the script to use the HTML5 doctype.
As you may recall, the script uses a per-directory TEMPLATE.html file. Those
can be changed to the HTML5 doctype one directory at a time as long as the
person
Upgrade complete. Please let me know if you have any trouble with the new
server.
-Bill
On Jan 18, 2011, at 6:50 AM, William Siegrist wrote:
> The trac and svn servers will be down from 7 to 8am PST on January 19th for a
> hardware upgrade. The git mirror will still be available during this
I ran into similar problem several times, it might be a DumpRenderTree
bug. One DumpRenderTree instance execute 1000 tests without restarting
by default. Sometimes it occurs that a test leaves behind some mess in
the DRT and it breaks one of the following tests. If the test passes
with "run-webkit
Two of the commit-queue bots are failing CSS font tests after
yesterday. That brings the commit-cluster down to one bot for the
time being.
If anyone has any theories, we're tracking the failure here:
https://bugs.webkit.org/show_bug.cgi?id=52708
Expect slightly-longer-than-normal delays (1-2 ho
11 matches
Mail list logo
