Nathan Froyd wrote:
If you have objections to this change, please speak up, either here or in bug
794178.
set NSPR_LOG_FILE=WinDebug
is currently the only way to get assertions and warnings logged in the
debugger output window. (Other output, such as dump(), is special-cased.)
--
Warning:
On 2013-01-03 1:28 PM, Benjamin Smedberg wrote:
On 1/3/13 1:23 PM, Boris Zbarsky wrote:
On 1/3/13 1:08 PM, Benjamin Smedberg wrote:
Do we know why this was happening? The log module specifies "nsDebug",
and it should only be written to the log if PR_LOG_MODULES includes
"nsDebug".
It's becaus
On 1/3/13 1:23 PM, Boris Zbarsky wrote:
On 1/3/13 1:08 PM, Benjamin Smedberg wrote:
Do we know why this was happening? The log module specifies "nsDebug",
and it should only be written to the log if PR_LOG_MODULES includes
"nsDebug".
It's because InitLog in nsDebugImpl does this:
gDebugLo
On 1/3/13 1:08 PM, Benjamin Smedberg wrote:
Do we know why this was happening? The log module specifies "nsDebug",
and it should only be written to the log if PR_LOG_MODULES includes
"nsDebug".
It's because InitLog in nsDebugImpl does this:
gDebugLog = PR_NewLogModule("nsDebug");
gDebu
On 1/2/13 9:11 PM, Boris Zbarsky wrote:
On 1/2/13 4:44 PM, Benjamin Smedberg wrote:
Was this particular logging enabled with NSPR_LOG_MODULES=all:5
No.
The failure mode was that if you enabled logging for any module at all
we would crash at shutdown because there was an NS_WARNING after
poi
On 1/2/13 4:44 PM, Benjamin Smedberg wrote:
Was this particular logging enabled with NSPR_LOG_MODULES=all:5
No.
The failure mode was that if you enabled logging for any module at all
we would crash at shutdown because there was an NS_WARNING after
poisoning happened (from NS_ENSURE_SOMETHING
- Original Message -
> There are potentially lots of ways to cause logging to happen after
> poisoning is enabled. I don't understand why we shouldn't instead
> whitelist logging as ok during shutdown.
The whitelist technique that has been used for other files, e.g.
whitelisting particular
On 1/2/2013 2:34 PM, Nathan Froyd wrote:
Hi all,
Over in bug 794178, we're hitting cases where the combination of write
poisoning (crashing if writes to disk are detected after a certain point) and
logging are causes crashes. These same crashes are blocking testing for WebRTC
(bug 811764).
8 matches
Mail list logo
