Also freeStackBuffer() and all the funky classic-C string management of a big
flat buffer divided into exactly 512 128-byte strings. Define StringVector as
a std::vector<std::string>, and use that instead.
Retain the behavior of clearing the vector if it exceeds 512 entries.
This eliminates the LLError::Log::flush(const std::ostringstream&, char*)
overload as well, with its baffling mix of std::string and classic-C (e.g.
strlen(out.str().c_str()).
If we absolutely MUST use a big memory pool for performance reasons, let's
use StringVector with allocators.
Opening and closing single-instance floaters rapidly and multiple times could result in cleaning instance from floaterreg twice, thus removing new isntance instead of current one
Instead of a single std::ostringstream instance shared by all callers, even
those on different threads, make each of the relevant lllog_test_() and
llcallstacks macros instantiate independent (stack) std::ostringstream
objects. lllog_test_() is called by LL_DEBUGS(), LLINFOS(), LL_WARNS(),
LL_ERRS(), LL_VLOGS() et al.
Eliminate LLError::Log::out(), whose sole function was to arbitrate use of
that shared std::ostringstream. Amusingly, if the lock couldn't be locked or
if messageStreamInUse was set, out() would allocate a new (heap!)
std::ostringstream anyway, which would then have to be freed by flush().
Make both LLError::Log::flush() overloads accept const std::ostringstream&.
Make LL_ENDL pass the local _out instance. This eliminates the need to check
whether the passed std::ostringstream* references the shared instance and
(if so) reset it or (if not) delete it.
Make LLError::LLCallStacks::insert() accept the local _out instance as non-
const std::ostream&, rather than acquiring and returning std::ostringstream*.
Make end() accept the local instance as const std::ostringstream&.
Introduce 'string_params' typedef for std::initialization_list<std::string>,
and make logwarns(), loginfos(), logdebugs() and logerrs() accept const
string_params&.
Eliminate the central log() function in llsingleton.cpp that used LL_VLOGS().
To cache the result of a (moderately expensive) Log::shouldLog() call,
LL_VLOGS() wants its CallSite object to be static -- but of course the
shouldLog() result will differ for different ELevel values, so LL_VLOGS()
instantiates a static array of CallSite instances. It seems silly to funnel
distinct logwarns(), etc., functions through a common log() function only to
have LL_VLOGS() tease apart ELevel values again. Instead, make logwarns()
directly invoke LL_WARNS(), and similarly for the rest.
To reduce boilerplate in these distinct functions, teach std::ostream how to
stream a string_params instance by looping over its elements. Then each
logwarns(), etc., function can simply stream its string_params argument to
LL_WARNS() or whichever.
In particular, eliminate the LLERROR_CRASH macro in logerrs(). The fact that
it invokes LL_ERRS() ensures that its LL_ENDL macro will crash the viewer.
Introduce Oz's LLERROR_CRASH macro analogous to the old LLError::crashAndLoop()
function. Change LL_ENDL macro so that, after calling flush(), if the CallSite
is for LEVEL_ERROR, we invoke LLERROR_CRASH right there.
Change the meaning of LLError::FatalFunction. It used to be responsible for
the actual crash (hence crashAndLoop()). Now, instead, its role is to disrupt
control flow in some other way if you DON'T want to crash: throw an exception,
or call exit() or some such. Any FatalFunction that returns normally will fall
into the new crash in LL_ENDL.
Accordingly, the new default FatalFunction is a no-op lambda. This eliminates
the need to test for empty (not set) FatalFunction in Log::flush().
Remove LLError::crashAndLoop() because the official LL_ERRS crash is now in
LL_ENDL.
One of the two common use cases for setFatalFunction() used to be to intercept
control in the last moments before crashing -- not to crash or to avoid
crashing, but to capture the LL_ERRS message in some way. Especially when
that's temporary, though (e.g. LLLeap), saving and restoring the previous
FatalFunction only works when the lifespans of the relevant objects are
strictly LIFO.
Either way, that's a misuse of FatalFunction. Fortunately the Recorder
mechanism exactly addresses that case. Introduce a GenericRecorder template
subclass, with LLError::addGenericRecorder(callable) that accepts a callable
with suitable (level, message) signature, instantiates a GenericRecorder, adds
it to the logging machinery and returns the RecorderPtr for possible later use
with removeRecorder().
Change llappviewer.cpp's errorCallback() to an addGenericRecorder() callable.
Its role was simply to update gDebugInfo["FatalMessage"] with the LL_ERRS
message, then call writeDebugInfo(), before calling crashAndLoop() to finish
crashing. Remove the crashAndLoop() call, retaining the gDebugInfo logic. Pass
errorCallback() to LLError::addGenericRecorder() instead of setFatalFunction().
Oddly, errorCallback()'s crashAndLoop() call was conditional on a compile-time
SHADER_CRASH_NONFATAL symbol. The new mechanism provides no way to support
SHADER_CRASH_NONFATAL -- it is a Bad Idea to return normally from any LL_ERRS
invocation!
Rename LLLeapImpl::fatalFunction() to onError(). Instead of passing it to
LLError::setFatalFunction(), pass it to addGenericRecorder(). Capture the
returned RecorderPtr in mRecorder, replacing mPrevFatalFunction. Then
~LLLeapImpl() calls removeRecorder(mRecorder) instead of restoring
mPrevFatalFunction (which, as noted above, was order-sensitive).
Of course, every enabled Recorder is called with every log message. onError()
and errorCallback() must specifically test for calls with LEVEL_ERROR.
LLSingletonBase::logerrs() used to call LLError::getFatalFunction(), check the
return and call it if non-empty, else call LLError::crashAndLoop(). Replace
all that with LLERROR_CRASH.
Remove from llappviewer.cpp the watchdog_llerrs_callback() and
watchdog_killer_callback() functions. watchdog_killer_callback(), passed to
Watchdog::init(), used to setFatalFunction(watchdog_llerrs_callback) and then
invoke LL_ERRS() -- which seems a bit roundabout. watchdog_llerrs_callback(),
in turn, replicated much of the logic in the primary errorCallback() function
before replicating the crash from llwatchdog.cpp's default_killer_callback().
Instead, pass LLWatchdog::init() a lambda that invokes the LL_ERRS() message
formerly found in watchdog_killer_callback(). It no longer needs to override
FatalFunction with watchdog_llerrs_callback() because errorCallback() will
still be called as a Recorder, obviating watchdog_llerrs_callback()'s first
half; and LL_ENDL will handle the crash, obviating the second half.
Remove from llappviewer.cpp the static fast_exit() function, which was simply
an alias for _exit() acceptable to boost::bind(). Use a lambda directly
calling _exit() instead of using boost::bind() at all.
In the CaptureLog class in llcommon/tests/wrapllerrs.h, instead of statically
referencing the wouldHaveCrashed() function from test.cpp, simply save and
restore the current FatalFunction across the LLError::saveAndResetSettings()
call.
llerror_test.cpp calls setFatalFunction(fatalCall), where fatalCall() was a
function that simply set a fatalWasCalled bool rather than actually crashing
in any way. Of course, that implementation would now lead to crashing the test
program. Make fatalCall() throw a new FatalWasCalled exception. Introduce a
CATCH(LL_ERRS("tag"), "message") macro that expands to:
LL_ERRS("tag") << "message" << LL_ENDL;
within a try/catch block that catches FatalWasCalled and sets the same bool.
Change all existing LL_ERRS() in llerror_test.cpp to corresponding CATCH()
calls. In fact there's also an LL_DEBUGS(bad tag) invocation that exercises an
LL_ERRS internal to llerror.cpp; wrap that too.
Instead of accepting a fixed list of (const char* p1="", etc.), accept
(std::initializer_list<std::string_view>). Accepting a
std::initializer_list<T> in your parameter list allows coding (e.g.)
func({T0, T1, T2, ... });
-- in other words, you can pass the initializer_list a brace-enclosed list of
an arbitrary number of instances of T.
Using std::string_view instead of const char* means we can pass *either* const
char* or std::string. string_view is cheaply constructed from either, allowing
uniform treatment within the function.
Constructing string_view from std::string simply extracts the pointer and
length from the std::string.
Constructing string_view from const char* (e.g. a "string literal") requires
scanning the string for its terminating nul byte -- but that would be
necessary even if the scan were deferred until the function body. Since
string_view stores the length, the scan still happens only once.
Andrey Kleshchev