On Thu, Jul 31, 2025 at 11:44:16AM +0200, Pavel Sanda wrote: > On Thu, Jul 31, 2025 at 02:37:06AM +0000, test...@tester.lyx.org wrote: > > CXX Trans.o > > Trans.cpp: In member function 'const lyx::docstring > > lyx::Trans::process(lyx::char_type, lyx::TransManager&)': > > Trans.cpp:365:40: error: '((const > > __gnu_cxx::__alloc_traits<std::allocator<wchar_t>, > > wchar_t>::value_type*)((char*)&t + offsetof(const std::__cxx11::docstring, > > std::__cxx11::basic_string<wchar_t, std::char_traits<wchar_t>, > > std::allocator<wchar_t> >::<unnamed>)))[1]' may be used uninitialized > > [-Werror=maybe-uninitialized] > > 365 | return k.deadkey(c, > > kmod_list_[static_cast<tex_accent>(t[1])]); > > | ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > > Trans.cpp:357:25: note: 't' declared here > > 357 | docstring const t = match(c); > > | ^ > > > I do not know what the code is supposed to do, but clearly we don't test for > the combination of > t.empty() && c = 0 > > Can such case appear? If so should we return docstring()?
It's probably better to assert than just access unallocated memory. I am tempted to commit the attached. Anyone opinion? Pavel
diff --git a/src/Trans.cpp b/src/Trans.cpp index 2a3daf9df0..12086658b1 100644 --- a/src/Trans.cpp +++ b/src/Trans.cpp @@ -17,6 +17,7 @@ #include "support/debug.h" #include "support/docstream.h" +#include "support/lassert.h" #include "support/FileName.h" #include "support/filetools.h" #include "support/Lexer.h" @@ -362,6 +363,8 @@ docstring const Trans::process(char_type c, TransManager & k) if (!t.empty() && t[0] != 0) return t; //return k.normalkey(c); + LASSERT(t.size()>1, return docstring()); + return k.deadkey(c, kmod_list_[static_cast<tex_accent>(t[1])]); }
-- lyx-devel mailing list lyx-devel@lists.lyx.org https://lists.lyx.org/mailman/listinfo/lyx-devel
