Amebis/wxWidgets
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Document wxNO_UNSAFE_WXSTRING_CONV in wxString overview

Browse Source 
Also add wxUSE_UNSAFE_WXSTRING_CONV to the list of wxUSE_XXX options
documentation and mention it in wxString overview as well.

See #11830.
This commit is contained in:
Vadim Zeitlin
2018-07-11 00:44:46 +02:00
parent f54ac4e243
commit 3dd4fb3997
2 changed files with 27 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
docs/doxygen/mainpages/const_wxusedef.h
View File

@@ -45,6 +45,9 @@ library:
other locales and reduces the library size; useful for embedded systems.} other locales and reduces the library size; useful for embedded systems.}
@itemdef{wxUSE_GUI, Use the GUI classes; if set to 0 only non-GUI classes are @itemdef{wxUSE_GUI, Use the GUI classes; if set to 0 only non-GUI classes are
available.} available.}
@itemdef{wxUSE_UNSAFE_WXSTRING_CONV, Defined as 1 by default for compatibility,
but can be set to 0 to globally disable unsafe implicit conversions in
wxString class. See @ref overview_string_settings for more details.}
@endDefList @endDefList

28
docs/doxygen/overviews/string.h
View File

@@ -178,10 +178,22 @@ which are deprecated and may disappear in future versions.
@subsection overview_string_implicitconv Implicit conversions @subsection overview_string_implicitconv Implicit conversions
Probably the main trap with using this class is the implicit conversion The default behaviour, which can't be changed to avoid breaking compatibility
operator to <tt>const char*</tt>. It is advised that you use wxString::c_str() with the existing code, is to provide implicit conversions of wxString to
instead to clearly indicate when the conversion is done. Specifically, the C-style strings, i.e. <tt>const char*</tt> and/or <tt>const wchar_t*</tt>. As
danger of this implicit conversion may be seen in the following code fragment: explained below, these conversions are dangerous and it is @e strongly
recommended to predefine @c wxNO_UNSAFE_WXSTRING_CONV for all new projects
using wxWidgets to disable them. Notice that this preprocessor symbol is
different from the more usual @c wxUSE_XXX build options, as it only needs to
be defined when building the application and doesn't require rebuilding the
library (and so can be used with e.g. system-provided libraries from Linux
system packages).
If you can't disable the implicit conversions, it is still advised to use
wxString::c_str() instead of relying on them to clearly indicate when the
conversion is done as implicit conversions may result in difficult to find
bugs. For example, some of the dangers of this implicit conversion may be seen
in the following code fragment:
@code @code
// this function converts the input string to uppercase, // this function converts the input string to uppercase,
@@ -374,6 +386,14 @@ using UTF-8 has performance implications as explained in @ref
overview_unicode_performance, so it probably shouldn't be enabled for legacy overview_unicode_performance, so it probably shouldn't be enabled for legacy
code which might contain a lot of index-using loops. code which might contain a lot of index-using loops.
As mentioned in @ref overview_string_implicitconv, @c wxNO_UNSAFE_WXSTRING_CONV
should be defined by all code using this class to opt-in safer, but not
backwards-compatible, behaviour of @e not providing dangerous implicit
conversions to C-style strings. This option is convenient when using standard
build of the library as it doesn't require rebuilding it, but for custom builds
it is also possible to set @c wxUSE_UNSAFE_WXSTRING_CONV to 0 in order to
disable the implicit conversions for all applications using it.
See also @ref page_wxusedef_important for a few other options affecting wxString. See also @ref page_wxusedef_important for a few other options affecting wxString.
*/ */