As the 2016-10-07~09 downtime incident shows, it is paramount to take
further steps in guaranteeing that the server can't corrupt its own data
files (especially the add-ons database) when receiving
inappropriately-timed signals.
This commit adds and deploys an ostream wrapper that requires callers to
explicitly commit the results to disk when finished writing to the
stream, so that only then the real destination file is overwritten with
the working copy (a temporary in the same dir). This way, there should
never be a situation in which the destination is left in an inconsistent
state due to a signal or exception.
The temporary receives a predictable name right now in the interest of
simplicity, since we are more or less in control of the target directory
anyway. We definitely don't want it to be an unlinked file since that
would make it impossible for admins to inspect and compare the temporary
against the original afterwards.
The code makes some assumptions about the nature of the return value of
filesystem::ostream_file() which will never be broken in this stable
branch, which is why one helper function is in campaignd land rather
than in the global filesystem API for now. This should probably be
rectified when forward-porting to master. Maybe.
Nothing of this will work reliably on Windows but we don't care. There's
only one machine in the world where we support running campaignd at this
time and it runs Linux.
this has 2 advantages:
1) we can now forwaed declare the class config_attribute_value without
including config.hpp
2) the config.?pp files are a little easier to read.
This also removes the included SHA1 and MD5 implementations
in favour of those from OpenSSL.
Thanks to @jyrkive for helping to get the CMake build working and
ensuring the game still compiles with OpenSSL 1.1.0.
(The SHA_xxx() aliases were removed in OpenSSL 1.1.0.)
The game now supports @ or = in usernames for the purpose of saving them to a file.
Though I don't think these are allowed currently, it's probably better to
support it from the start rather than having things break if they later
become allowed.
Looking at this more I realized the help_button was a GUI1 dialog component meant to open the Help
dialog. Since all dialogs invoking Help now use GUI2, there's no need for this class anymore. It
also allows a bunch of remaining GUI1 stuff to be cleaned up.
This was a dialog used early in 1.13 during the SP/MP codepath unification to configure certain options.
The modification selection is now built into the Selection dialog, and we decided not to enable other options
like Era selection in SP. The only option we might look to restoring in the future is the "show_connect"
option, but that requires some additional code elsewhere and probably won't be used anyway.
I left the controller passed in an argument for compatibility with the GUI1 dialog, but since that's removed
it makes sense that the dialog handle the creation of the controller, especially since due to an oversight
it made a copy instead of holding a reference.
This also allows storyscreen/interface.*pp to be removed. The START_POSITION was part of the GUI1 implementation
and isn't needed anymore.
This splits the handling of each applicable type into its own helper class to better encapsulate the functionality of each.
It also contains a lot of related cleanup and improvements to the class interface itself.
The purpose of this change is ability to create more dialogs which present
a list of add-ons, such as a GUI2 version of the "Install Dependencies"
dialog.
This change is still highly incomplete. First of all, I disabled inline
install buttons and ability to sort the add-on list. In addition, a lot of
functionality that belongs in the add-on list widget is still in the GUI2
add-on manager instead. It is also notable that the widget relies on
several linked groups being defined in the window.
I plan to finish the transition to the add-on list widget, and address the
problems in the previous paragraph, in the future. The linked group issue
is going to require a lot of work in particular: I plan to either make
linked groups grid-specific instead of window-specific (like they are now)
or allow widgets to inject linked groups to the containing window. Either
way, I'm going to change the operation of GUI2 fairly deeply.
* Use separate files to build scons and to build lua for cmake
Move lists of source files into their own directory
Make scons use the lists of source files
Move lua build step to src/SConscript and delete src/lua/SConscript
Make cmake use the separate file with the list of lua sources to build
* Finish changing over to source lists
scons
move source files from libraries that don't appear in cmake over to existing source lists - libcampaignd, libcutter, libdummy_video, libtest_utils
add the defines FIFODIR and WESNOTH_PATH to all compiled source files, as cmake does, and move the single files those defines had previously been defined on into the source lists
remove WESNOTH_PREFIX, as not used
added schema_validator sources from cmake
added OBJPREFIX to - cutter, exploder, campaignd, schema_generator, schema_validator
cmake
change over to using the source lists
move source files from libraries that don't appear in scons over to existing source lists - wesnoth-gui_types, wesnoth-gui_event, wesnoth-gui_iterator, wesnoth-gui_placer, wesnoth-gui_tooltip, wesnoth-gui_widget, wesnoth-gui_widget_definition, wesnoth-gui1_widgets, wesnoth-schema_validator
