r=bryner
1) Solaris should use the same option for |find| with Darwin;
2) Use `which foo | grep /foo$` instead of `which foo`, because Solaris |which| command always outputs something even the command is not found.
- show progress while loading the events file
- show more detailed leak summary
no review, this is a debug-only tool and I'm only changing console output
r=bernd
This is dreftool, resurrected with the help of kherron, bernd, and myself.
Licensing has been switched per http://bugzilla.mozilla.org/show_bug.cgi?id=8227#c45
to MPL/GPL/LGPL.
Old copies of xpcom classes have been removed, dreftool now uses the normal ones.
New:
* Builds on Linux and uses NSPR, so it should be possible to use it on most
systems.
* Member variables are screened.
* Tries to recognize NS_ENSURE early outs.
* Can parse a single file.
* The summary can indicate what it was told to parse.
* Links to LXR can be generated. Using this feature is not recommended as pieces of
mozilla tend to be on branches and the links will be bad. If for non branched content,
the links will drift as the files are changed.
* Links to Bonsai against the file revision will be generated if CVS records are
detected. This is good, unless you have local modifications in your tree.
Known issues:
* Aliasing isn't detected, so
|a = new something; *foo = a; if (!*foo) return NS_ERROR_FAILURE;| is flagged.
e.x.
http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/accessible/src/html/nsHTMLSelectAccessible.cpp&rev=1.33&mark=920-921,922,923,925#920
* Assigning to dereferenced objects
|*foo = new something| is not quite correctly handled (perhaps it's misinterpreted as
|foo = new something|. e.x.
http://bonsai.mozilla.org/cvsblame.cgi?file=/mozilla/extensions/transformiix/source/xslt/txStandaloneXSLTProcessor.cpp&rev=1.11&mark=88,92,96,99,84#91
* Various other allocators aren't caught: strdup and co., nsCRT::, nsMemory::, ...
* There's no way to provide a user defined list of allocators.
People should of course not file bugs simply based on dreftool, they should file bugs
after verifying that the complaints from dreftool are valid.
r=bernd
This is dreftool, resurrected with the help of kherron, bernd, and myself.
Licensing has been switched per http://bugzilla.mozilla.org/show_bug.cgi?id=8227#c45
to MPL/GPL/LGPL.
Old copies of xpcom classes have been removed, dreftool now uses the normal ones.
New:
* Builds on Linux and uses NSPR, so it should be possible to use it on most
systems.
* Member variables are screened.
* Tries to recognize NS_ENSURE early outs.
* Can parse a single file.
* The summary can indicate what it was told to parse.
* Links to LXR can be generated. Using this feature is not recommended as pieces of
mozilla tend to be on branches and the links will be bad. If for non branched content,
the links will drift as the files are changed.
* Links to Bonsai against the file revision will be generated if CVS records are
detected. This is good, unless you have local modifications in your tree.
Known issues:
* Aliasing isn't detected, so
|a = new something; *foo = a; if (!*foo) return NS_ERROR_FAILURE;| is flagged.
e.x.
http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/accessible/src/html/nsHTMLSelectAccessible.cpp&rev=1.33&mark=920-921,922,923,925#920
* Assigning to dereferenced objects
|*foo = new something| is not quite correctly handled (perhaps it's misinterpreted as
|foo = new something|. e.x.
http://bonsai.mozilla.org/cvsblame.cgi?file=/mozilla/extensions/transformiix/source/xslt/txStandaloneXSLTProcessor.cpp&rev=1.11&mark=88,92,96,99,84#91
* Various other allocators aren't caught: strdup and co., nsCRT::, nsMemory::, ...
* There's no way to provide a user defined list of allocators.
People should of course not file bugs simply based on dreftool, they should file bugs
after verifying that the complaints from dreftool are valid.
- Give example conf file for mod_perl 2 in README.txt
- Use RegistryPrefork handler to emulate mod_perl 1.x's chdir() behavior
- Add version of RegistryPrefork.pm backported for mod_perl <= 1.99_08
- Don't autoflush output in the middle of sending HTTP headers; it confuses mod_perl's ParseHeaders
- Comment out a warn() that's not really needed
r=jrgm
Remove the hardcoded module2dir mapping and use the results from echo-requires-recursive instead.
Add README which briefly explains how module-deps is supposed to work.
Add XSLT hacks to modules.mk.footer.
Add cygdrive hacks to modules.mk.stub.
Force string & embedstring to be built after xpcom since they now live in the xpcom hierarchy.
Remove the hardcoded module2dir mapping and use the results from echo-requires-recursive instead.
Add README which briefly explains how module-deps is supposed to work.
Add XSLT hacks to modules.mk.footer.
Add cygdrive hacks to modules.mk.stub.
Force string & embedstring to be built after xpcom since they now live in the xpcom hierarchy.
