Currently, creating a StringVal from a std::string did not work with
data that contains \0 characters. This easy fix changes this - and
should also lead to a small speed increase for code using this
constructor.
This obviously means that more data might copied now in some cases that
were previously cut off at the first 0-byte. Our test-suite did not
reveal any such cases.
The default value of an ID is now truly the one used to initialize it,
unaltered by any subsequent redefs.
Redefs are now shown separately, along with the expression that
modifies the ID's value.
* 'paraglob' of https://github.com/ZekeMedley/zeek:
Add leak test to paraglob.
Catch paraglob serialization errors in DoClone.
Update paraglob serialization.
Stop execution on paraglob error.
Update paraglob submodule
Change C++11 detection in paraglob.
Make paraglob serializable and copyable.
Initial paraglob integration.
I made a bunch of small changes:
* paraglob now deals better with \0 characters
* I rolled back the changes to Binary Serialization format,
* there were some small formatting issue
* the error output was slightly unsafe
* build_unique is now in util.h.
and perhaps a few more small things.
Was possible for some failed parsing calls in
RPC_Interpreter::DeliverRPC to set buf to a null-pointer, only for it
to later access it when checking whether the message length appeared
to be legit.
Adding the null-initialization of buf in RPC_Interpreter::Timeout()
was just precautionary -- didn't look like any real bugs with anyone
accessing its uninitialized state at the moment.
* origin/topic/timw/171-deprecation-warnings:
GH-171: support warning messages alongside deprecated attributes
Made a minor tweak to give a better error message when using
&deprecated= with something other than a string literal expression.
* 'rdp_client_cluster_data' of https://github.com/neslog/zeek:
Fixing types.
Correcting types.
Removing misc data from Client Cluster data trying to assign values.
Adding options field to RDP::ClientChannelDef Adding Client Cluster Data
Adjustments:
- Reformatting
- Added comments
- Changed the REDIRECTED_SESSIONID_FIELD_VALID field to a bool
Fixed a few small bugs - Modifiable had an uninitialized member and the
Registry looped over a map while deleting elements from it.
Fixes GH-319
* remotes/origin/topic/robin/gh59-when:
Renaming src/StateAccess.{h,cc} to src/Notifier.{h,cc}.
Clean up new code.
Remove MutableVal class.
Redo API for notifiers.
Remove most of MutableVal (but not the class itelf yet)
Remove enum Opcode.
Remove StateAccess class.
Redo NotfifierRegistry to no longer rely on StateAccess.
Add new test for when-statement watching global variables.
Couple of compile fixes.
* origin/topic/johanna/remove-serializer:
Fix memory leak introduced by removing opaque of ocsp_resp.
Change return value of OpaqueVal::DoSerialize.
Add missing ShallowClone implementation for SetType
Remove opaque of ocsp_resp.
Remove remnants of event serializer.
Fix cardinalitycounter deserialization.
Smaller compile fixes for the new opaque serialization.
Reimplement serialization infrastructure for OpaqueVals.
Couple of compile fixes.
Remove const from ShallowClone.
Remove test-case for removed functionality
Implement a Shallow Clone operation for types.
Remove value serialization.
Various changes I made:
- Fix memory leak in type-checker for opaque vals wrapped in broker::data
- Noticed the two "copy-all" leak tests weren't actually checking for
memory leaks because the heap checker isn't active until after zeek_init()
is evaluated.
- Change OpaqueVal::DoClone to use the clone caching mechanism
- Improve copy elision for broker::expected return types in the various
OpaqueVal serialize methods
- Not all compilers end up properly treating the return of
local/automatic variable as an rvalue that can be moved, and ends up
copying it instead.
- Particularly, until GCC 8, this pattern ends up copying instead of
moving, and we still support platforms whose default compiler
pre-dates that version.
- Generally seems it's something that wasn't addressed until C++14.
See http://www.open-std.org/jtc1/sc22/wg21/docs/cwg_defects.html#1579
- Change OpaqueVal::SerializeType to return broker::expected
- Change probabilistic DoSerialize methods to return broker::expected
Since people forgetting to checkout submodules is such a common failure
case - update configure to give an error message is the cmake directory
seems to be missing.
This just checks for the presence of cmake/COPYING when a .git directory
is found; if cmake/COPYING is not present an error message is displayed.
* origin/topic/timw/393-vector-slicing:
Fix memory leak in vector slice assignment
Misc. tweaks to vector slicing implementation
Add ability to grow/shrink a vector using slicing, also adds Insert/Remove methods for VectorVal
Allow assignment for vectors using slices
Check for integral slice indexes, add extra test for [:]
Return an empty vector if the indices for slicing don't make sense
GH-393: Add slice notation for vectors
Two parts to this:
* Only allow vector slice assignment in statement contexts, not in
arbitrary assignment expressions. E.g. it's not clear what the
resulting value of `(v[1:2] = vector(1))` is for further expression
chaining. For reference, Python doesn't allow it either.
* Add a subclass of AssignExpr to specialize the behavior for index
slice assignments (because its behavior regarding expression
chaining is different per the previous point) and Unref the RHS
of things like `v[1:2] = vector(1)` after IndexExpr::Assign is
finished inserting it (since no one else takes ownership of it).
Instead of using an Expr subclass, IndexSliceAssignExpr, we could
use a proper Stmt, since that's the only context we currently use it
for, but if we did ever to decide on allowing its use in arbitrary
expression contexts, then I expect we'll need it this way anyway
(just with a different IndexSliceAssignExpr::Eval implementation).
* Minor style/format changes
* Fix a signed/unsigned comparison compiler warning
* Use a non-fatal error for non-integral slice indices so we can
report any further scripting errors instead of stopping the parse
right there
Now it returns a broker::expected<broker::data>, instead of directly
returning a broker::data before. This means that broker::data does no
longer have to be abused to convey error information.
In a way it would be kind of neat to have more fine-granular broker error
types for this use-case - at the moment everything returns
broker::ec::invalid_data, which seems to be the only reasonable choice.
It turns out that SetType was missing a ShallowClone implementation.
Which meant that when a SetType was cloned, a TableType was received
(which has one less member). Which resulted in invalid memory accesses
when using the clone.
Thank you valgrind :)
