also some improvements to the plugin
Change-Id: I0e3a519d70756e577fcb1bd47dd66864b5b4c871
Reviewed-on: https://gerrit.libreoffice.org/23289
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
... as shown by 7da15debe39b50255cc08b3fef7ae38185a9174f.
Which means that MenuManager is unused, and MenuDispatcher
does nothing useful.
Change-Id: Ic85619531054a573d971d38b52609e1cbcaf780c
Reviewed-on: https://gerrit.libreoffice.org/23117
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
and teach the plugin about code that takes the address of a function
Change-Id: Ia9d5afef44520aca236659e8176f1e27135ef4fc
Reviewed-on: https://gerrit.libreoffice.org/22861
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
...instead of trying to second-guess what to strip off $(CXX) to make it fit.
Keep the old way for existing build scripts that rely on it.
Change-Id: I145bdcba6d02002a9b653b4deb6e7f5a9c76cc8f
This reverts commit 4101fa184150663ddee6688c19eb4a44e341e8d4.
Just configure --without-parallelism and always use an explicit make -jN, and be
done with it. I just can't stand that "make[1]: Entering/Leaving directory"
noise around each "[build CXX] compilerplugins/clang/*.cxx" line any more.
A ridiculously fast way of doing this is:
for i in $(pcregrep -l -M -r --include='.*[hc]xx$' \
--exclude-dir=workdir --exclude-dir=instdir '^
{3,}' .)
do
perl -0777 -i -pe 's/^
{3,}/
/gm' $i
done
Change-Id: Iebb93eccbee9e4fc5c4380474ba595858a27ac2c
Reviewed-on: https://gerrit.libreoffice.org/22224
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Chris Sherlock <chris.sherlock79@gmail.com>
...so that isZeroConstant doesn't trigger an assert inside Clang's
isCXX11ConstantExpr when expr is sizeof(x) with x being dependent on a template
argument.
Change-Id: I6bab46e64cc085d597db25994d8bfdc66417fe83
Find code that compares floating point values with == or !=
It should rather use rtl::math::approxEqual
Change-Id: I9026e08823340fa1d6a042c430515344c93215bd
Reviewed-on: https://gerrit.libreoffice.org/21997
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
generates a warning about subclasses that add members but do not
override operator==()
Change-Id: If6df1a2cbd115f17bcca22f9b7995181dcf55c03
Reviewed-on: https://gerrit.libreoffice.org/20468
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Michael Stahl <mstahl@redhat.com>
Offline discussion about tdf#96067 "Crash on undo row inserts" brought up the
idea to warn about cases where uses of dynamic_cast are statically knwon to
always fail. Clang's clang::AST::CXXDynamicCastExpr::isAlwaysNull already
implements such a check, reporting true if the casted-from class is final, but
has two issues:
For one, it does not work for template code, when one of the involved types is a
template parameter type (so e.g., DestType->castAs<PointerType>() can crash).
For another, it misses the opportunity to report true if the casted-to type is
final and only derives from the casted-from type non-publicly. My hope was that
this, after the "final" decorations in 548c43238d02b34cf73e7c2ca1a912ee4fe82544
"Mark some classes as final," might turn up the culprit of tdf#96067 (with a
scenario similar to the failed dynamic_cast on private derivation in
63b67ab5cab8cf7576a68cabe5fb1a42c6ad800c "Use public derivation, and remove
then-unnecessary downcasts")---but not so.
Change-Id: I962ee19820758f9c601f4a292da7f37fa9dff5ce
...to flag places that implicitly derive a class privately instead of publicly,
where accidental private derivation can cause unexpected failure of
dynamic_cast, cf. 63b67ab5cab8cf7576a68cabe5fb1a42c6ad800c "Use public
derivation, and remove then-unnecessary downcasts."
Change-Id: I4bcd5c79c7e27380c820e2dd072fa4c4e8980078
