This code was skipping fsync on BTRFS because of an old guarantee about
the overwrite-by-rename behavior that no longer holds true. This has
been confirmed by the BTRFS developers to no longer be guaranteed since
Kernel 3.17 (August 2014), but it was guaranteed when this optimization
was first introduced in 2010.
This could result in empty files after crashes in applications using
g_file_set_contents(). Most prominently this might have been the cause
of dconf settings getting lost on BTRFS after crashes due to the
frequency with which such writes can happen in dconf.
See: https://gitlab.gnome.org/GNOME/dconf/-/issues/73
This is a follow-up from commit 995823b9d9, which added the condition
```
```
to the array test. On most platforms, both of those symbols are numeric
literals, but on 64-bit Windows `G_MAXSIZE` includes some widening
casts, which means it can’t be used in a preprocessor condition.
We don’t expose an appropriate symbol in `glibconfig.h` which could be
used instead, but the standard `*_WIDTH` symbols from `limits.h` will be
identical and can be used instead.
Signed-off-by: Philip Withnall <pwithnall@endlessos.org>
Fixes: #2565
This commit puts an additional underscore before the external symbol
`_g_binary_test1_resource_data` when using gcc for Windows, to match
the compiler's expectation.
Fixes#2571
Signed-off-by: Ralf Habacker <ralf.habacker@freenet.de>
Using ld_flags would work, but that does not propagate ldflags to users
of glib. Meson's dependency() call will propagate apple framework
dependencies to downstream users.
Fix for commit 20c8ea1bc651bc4b79d39b80d42b468f6e7a2dc8; while Linux
seems happy to return the value for the invalid key we set above, BSD
returns NULL (which is probably a more valid thing to do).
Accept both.
Signed-off-by: Philip Withnall <pwithnall@endlessos.org>
Many UCRT (and msvcrt/msvcxx) functions open dialog boxes
by default for .... some reason. This is a problem because a test runner
waiting on a process to exit won't see it exit unless someone actually
clicks away the box, which won't happen on a CI machine.
Additionally g_abort unconditionally raises a debugging exception,
which, if uncaught, will cause windows error reporting to pop a dialog
Resolve the first problem by calling platform specific (but documented)
functions to change the CRT's behavior in g_test_init
Resolve the second by only throwing a debug exception if we're under
debugging, and just calling abort() otherwise.
This reduces the number of popups triggerd by `meson test` from
over 10 to about three on my machine, mostly in the spawn test code.
This warning warns on widening casts from integer to pointer type,
because those casts can be problematic for porting to new pointer
sizes. The code in question didn't do any bad things here so an
intermediate cast to size_t was added to silence the warning
This test was exploiting unspecified behavior w.r.t. the address of string
literals, It expected them to be pooled (the same literal has the same
address, at least within a TU), but MSVC does not pool by default,
leading to a failure.
