Historically, a 'when' condition performed an AST-traversal to locate
any index-expressions like `x[9]` and evaluated them so that it could
register the associated value as something for which it needs to receive
"modification" notifications.
Evaluating arbitrary expressions during an AST-traversal like that ignores
the typical order-of-evaluation/short-circuiting you'd expect if the
condition was evaluated normally, from its root expression.
Now, a new subclass of IndexExpr is used to keep track of all IndexExpr
results in the context of evaluating a 'when' condition without having
to do a secondary AST-traversal-and-eval. i.e. the first evaluation of
the full 'when' condition follows the typical expression-evaluation
semantics (as always), but additionally now captures all the values
a Trigger needs to monitor for modifications.
* origin/topic/timw/nullptr:
The remaining nulls
plugin/probabilistic/zeekygen: Replace nulls with nullptr
file_analysis: Replace nulls with nullptr
analyzer: Replace nulls with nullptr
iosource/threading/input/logging: Replace nulls with nullptr
Only 1% build time speedup, but still, it declutters the headers a bit.
Before this patch:
2565.17user 141.83system 2:25.46elapsed 1860%CPU (0avgtext+0avgdata 1489076maxresident)k
72576inputs+9130920outputs (1667major+49400430minor)pagefaults 0swaps
After this patch:
2537.19user 142.94system 2:26.90elapsed 1824%CPU (0avgtext+0avgdata 1434268maxresident)k
16240inputs+8887152outputs (1931major+48728888minor)pagefaults 0swaps
The Zeek code base has very inconsistent #includes. Many sources
included a few headers, and those headers included other headers, and
in the end, nearly everything is included everywhere, so missing
#includes were never noticed. Another side effect was a lot of header
bloat which slows down the build.
First step to fix it: in each source file, its own header should be
included first to verify that each header's includes are correct, and
none is missing.
After adding the missing #includes, I replaced lots of #includes
inside headers with class forward declarations. In most headers,
object pointers are never referenced, so declaring the function
prototypes with forward-declared classes is just fine.
This patch speeds up the build by 19%, because each compilation unit
gets smaller. Here are the "time" numbers for a fresh build (with a
warm page cache but without ccache):
Before this patch:
3144.94user 161.63system 3:02.87elapsed 1808%CPU (0avgtext+0avgdata 2168608maxresident)k
760inputs+12008400outputs (1511major+57747204minor)pagefaults 0swaps
After this patch:
2565.17user 141.83system 2:25.46elapsed 1860%CPU (0avgtext+0avgdata 1489076maxresident)k
72576inputs+9130920outputs (1667major+49400430minor)pagefaults 0swaps
- Adds new trigger namespace
- Adds trigger::Manager class as a new IOSource for keeping track of triggers and integrating them into the loop. Previously the loop relied on the event manager Drain() method to process all triggers on every loop, but now that the loop actively waits for events to occur, triggers would not fire when they needed to. Adding them as part of the loop ensures they're checked.
There's now an notifier::Modifiable interface class that class
supposed to signal modifications are to be derived from. This takes
the place of the former MutableValue class and also unifies how Val
and IDs signal modifications.
Scripting errors/mistakes now consistently generate a runtime error
which have the behavior of unwinding the call stack all the way out of
the current event handler.
Before, such errors were not treated consistently and either aborted
the process entirely or emitted a message while continuing to execute
subsequent statements without well-defined behavior (possibly causing
a cascade of errors).
The previous behavior also would only unwind out of the current
function (if within a function body), not out the current event
handler, which is especially problematic for functions that return
a value: the caller is essentially left a mess with no way to deal
with it.
This also changes the behavior of the startup/initialization process
to abort if there's errors during bro_init() rather than continue one
to the main run loop. The `allow_init_errors` option may change this
new, default behavior.
- 'when' statements were problematic when used in a function/event/hook
that had local variables with an assigned function value. This was
because 'when' blocks operate on a clone of the frame and the cloning
process serializes locals and the serialization of functions had an
infinite cycle in it (ID -> BroFunc -> ID -> BroFunc ...). The ID
was only used for the function name and type information, so
refactoring Func and subclasses to depend on those two things instead
fixes the issue.
- 'return when' blocks, specifically, didn't work whenever execution
of the containing function's body does another function call before
reaching the 'return when' block, because of an assertion. This was
was due to logic in CallExpr::Eval always clearing the CallExpr
associated with the Frame after doing the call, instead of restoring
any previous CallExpr, which the code in Trigger::Eval expected to
have available.
- An assert could be reached when the condition of a 'when' statement
depended on checking the value of global state variables. The assert
in Trigger::QueueTrigger that checks that the Trigger isn't disabled
would get hit because Trigger::Eval/Timeout disable themselves after
running, but don't unregister themselves from the NotifierRegistry,
which keeps calling QueueTrigger for every state access of the global.
