| Age | Commit message (Collapse) | Author |
|---|
|
ItemType is useful and faster than RTTI. Until now it was
implemented by a 16-bit member in the base class, plus
(potentially) all constructors having to hand a value
in at item construction type (of type SfxItemType) to
get that member set correctly.
This works, but there is no reliable way to guarantee
coverage, and there have already been cases with missing
SfxItemType - these fallback to '0' and thus all Items
with ItemType() == 0 are assumed equal and might be
static_cast'ed to the wrong classes. Note that I
identified *35* Items that had no correct ItemType
set/implemented actually. It also uses 16-bit per
incarnated Item at runtime.
I thought and realized now a more systematic approach
to do that with a pure virtual function at the Item
itself. That can also be secured by a clang compiler
plugin in the future to keep it working. It uses one
virtual function per derived class, no longer space
in incarnated Items. Also the constructors will get
more simple again.
But the main aspect is security - we cannot afford
Items potentially being held as equal if they are not.
Unfortunately C++ does not offer something like a
'strict pure virtual function' that would force to
be overloaded in every derivation, but the used
methotology and adding a clang test is reasonably
safe.
Have now done the cleanup of previous method.
Change-Id: I04768285f1e9b73d64b0bb87df401944b5d35678
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/180017
Tested-by: Jenkins
Reviewed-by: Armin Le Grand <Armin.Le.Grand@me.com>
|
|
This reverts commit 85fd526fc681a994415bb422090d1d23aa7d54f6.
Change-Id: I5019f72f88497f50a77666d57f2d16c2749bd1c9
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/170218
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
|
|
The mechanism is currently broken because it uses hash values as keys,
in two different places. But hash values are not required to be unique,
and as soon as there are enough items of a given type, a collision is
inevitable.
So just simplify this whole mechanism. There is no reason we need
type-specific item managers. Stuff everything into a single global pool,
that uses hashing correctly.
Notes
(*) Performance, as far as I can tell, is the same or slightly better.
(*) I removed the NUMBER_OF_UNSHARED_INSTANCES thing, in favour
of just having a simpler mechanism
Change-Id: I9068baf9bf6fae2500ae5748c6d1aabe6c3a18a4
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/169709
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
Tested-by: Jenkins
|
|
and can then remove some casting
Change-Id: Id821c32ca2cbcdb7f57ef7a5fa1960042e630ffc
Reviewed-on: https://gerrit.libreoffice.org/85022
Tested-by: Jenkins
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
|
|
Change-Id: I0814e98c82822752dd4410d595d35acb67d5a1d1
Reviewed-on: https://gerrit.libreoffice.org/81703
Tested-by: Jenkins
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
|
|
(1) Migrated all still existing binary load/save stuff
in SfxPoolItem to legacy files. Isolated from Item
implementations. Adapted all usages. No more methods
Create/Store needed, also GetVersion removed
(2) Removed operator= for SfxPoolItem. Adapted all
usages. Goal ist to handle Items more as Objects
('Object-Oriented') in the sense to move/handle
instances, not to copy one instance over another one
(which is more and more problematic with hard to copy
content as UNO API stuff or similar). This lead to
much more usages of std::shared_ptr which correlates
well with future plans fr Items (see dev branch).
Next logic step will be to also remove copy constructor
Linux build and corrections done
Fixed Writer test and removed unused defines
Fixed another unused m,acro
Started to unify the AutoFormat stuff
Changes to OUString constructor usages, tests completely
No idea why, but SfxStringItem constructor which
takes a OUString& now insists of not getting ::OUString's
handed in - changed all 'SfxStringItem.*OUString.*".*"'
accordingly
Change-Id: Ibed7358b18fb019994a7490332b9d797a6694c29
Reviewed-on: https://gerrit.libreoffice.org/71075
Tested-by: Jenkins
Reviewed-by: Armin Le Grand <Armin.Le.Grand@me.com>
|
|
This mostly affects explicitly defaulted functions that had recently been user-
declared to silence new GCC trunk -Wdeprecated-copy. It is not entirely clear
to me what the best approach is overall, see my mail
<http://lists.llvm.org/pipermail/cfe-commits/Week-of-Mon-20181001/245321.html>
"Re: r343285 - [cxx2a] P0641R2: (Some) type mismatches on defaulted functions
only", but lets just explicitly delete those functions for now.
Change-Id: If8c72f612f67a8feb8b03c2fb988c807e704ef03
Reviewed-on: https://gerrit.libreoffice.org/61259
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
...by explicitly defaulting the copy/move functions (and, where needed in turn,
also a default ctor) for classes that have a user-declared dtor that does
nothing other than an implicitly-defined one would do, but needs to be user-
declared because it is virtual and potentially serves as a key function to
emit the vtable, or is non-public, etc.
Change-Id: I8c900426c49701afc94094115ec0c2d5b1af90d9
Reviewed-on: https://gerrit.libreoffice.org/58102
Tested-by: Jenkins
Reviewed-by: Stephan Bergmann <sbergman@redhat.com>
|
|
SfxPoolItem has a copy constructor and its copy assignment
operator is deleted. Derived classes have a implicit defined
copy constructor too, if all members are copy constructible.
This patch removes default able copy constructors on such items.
Note: clang-tidy only finds functions defined in source, inlined
functions in headers are picked manualy.
Change-Id: I7e642a2ecf139420c53f6fcb39208918dd1a8d52
Reviewed-on: https://gerrit.libreoffice.org/43003
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
|
|
which requires explicitly adding null in
1) SdrItemBrowserControl::SetAttributes(const SfxItemSet* pSet, const SfxItemSet* p2ndSet)
where SdrItemBrowserControl is only used by SdrItemBrowser and the only use of that is
within DBG_UTIL in SdrPaintView
2) SwCursorShell::GetContentAtPos( const Point& rPt,
within a #ifdef DBG_UTIL block in
3) SvxSearchDialog::BuildAttrText_Impl( OUString& rStr,
bool bSrchFlag ) const
where the other branch uses SvxResId
4) SfxPoolItem::dumpAsXml(xmlTextWriterPtr pWriter) const
5) XFillStyleItem::dumpAsXml(xmlTextWriterPtr pWriter) const
looks very much like all uses (outside the dumpers) are intended
to be in the ui locale
results in that INetContentTypes::GetPresentation always called with UI Locale
Change-Id: I5a110c107838b4db3c355476426d6532f2b6ce52
Reviewed-on: https://gerrit.libreoffice.org/40538
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
|
|
Change-Id: Idb022a4a6fb950f1b259abbba57daed9401732d9
Reviewed-on: https://gerrit.libreoffice.org/31038
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noel.grandin@collabora.co.uk>
|
|
The issue of 362d4f0cd4e50111edfae9d30c90602c37ed65a2 "Explicitly mark
overriding destructors as 'virtual'" appears to no longer be a problem with
MSVC 2013.
(The little change in the rewriting code of compilerplugins/clang/override.cxx
was necessary to prevent an endless loop when adding "override" to
OOO_DLLPUBLIC_CHARTTOOLS virtual ~CloseableLifeTimeManager();
in chart2/source/inc/LifeTime.hxx, getting stuck in the leading
OOO_DLLPUBLIC_CHARTTOOLS macro. Can't remember what that
isAtEndOfImmediateMacroExpansion thing was originally necessary for, anyway.)
Change-Id: I534c634504d7216b9bb632c2775c04eaf27e927e
|
|
and
coverity#705367 Mixing enum types
coverity#705371 Mixing enum types
coverity#982694 Mixing enum types
coverity#1027717 Mixing enum types
coverity#1371228 Mixing enum types
coverity#1371242 Mixing enum types
coverity#1371280 Mixing enum types
coverity#1371310 Mixing enum types
MapUnit and SfxMapUnit share the same values and
are freely cast from one to the other.
Now that
commit d30a4298bdb5ba53cd1fe659f2b742f218a2e527
Date: Thu Aug 11 15:02:19 2016 +0200
loplugin:unusedenumconstants in package..svtools
removed the SfxMapUnit entries that were directly unused, they
don't match anymore and casting from one to the other is dangerous.
Why there was two of these anyway escapes me, get rid of SfxMapUnit
and just use MapUnit universally
Change-Id: I4db5dcd04b59be2f85b62b728f96c90afe00c57e
Reviewed-on: https://gerrit.libreoffice.org/28234
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Caolán McNamara <caolanm@redhat.com>
Tested-by: Caolán McNamara <caolanm@redhat.com>
|
|
Change-Id: I0f2f81dae91f62639e79799b1bed1b2df1fd79ab
|
|
Change-Id: I9cd92b53370a6b6018d2f7c648890f9c014a27de
|
|
Change-Id: I2ea407acd763ef2d7dae2d3b8f32525523ac8274
|
|
more useful to make it explicit.
Specifically, otherwise my defaultvalue clang plugin would want to
remove lots of places that contains #define constants which, while
technically the same as the default value, convey semantic
information which is quite useful.
Change-Id: I918ad5b0e73ba279fa1b1489b62d900339ff71eb
|
|
Change-Id: If86cc43fda4d138cf7f678d81fa2b35f68f3c03b
Reviewed-on: https://gerrit.libreoffice.org/17162
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
|
|
Change-Id: Ic136cce6abef44291b7236a6d709f0eee391f311
Reviewed-on: https://gerrit.libreoffice.org/16784
Tested-by: Jenkins <ci@libreoffice.org>
Reviewed-by: Noel Grandin <noelgrandin@gmail.com>
Tested-by: Noel Grandin <noelgrandin@gmail.com>
|
|
Change-Id: I5a01162d2925eede97f1cdc24aa876179b8b43b4
|
|
since all two of the actual call-sites only care about whether
it is a valid presentation or not, not what kind of presentation it is.
Change-Id: I75717c88878d37b2897741b0c833ff283b3fee59
|
|
Change-Id: I42119f656ca528286fb25d2d36c0af54b7d04a6b
|
|
...mostly done with a rewriting Clang plugin, with just some manual tweaking
necessary to fix poor macro usage.
Change-Id: I71fa20213e86be10de332ece0aa273239df7b61a
|
|
0.75 pt provides a better interop with Excel documents.
Change-Id: Ic1d2cbbe4e35dc0793a0e35d3836261d91138f7c
|
