Problem: on Windows, Esc does not cancel metadata editor.
Fix:
1. In tag.h, for windows, don't override IsEscapeKey().
2. In tag.cpp, for windows, include a return in the function DoCancel, so that if a cell is being edited and the user presses Esc, this just cancels the edit, and not the dialog.
This fix returns the behaviour of the dialog on Windows back to the behaviour prior to commit c1ca055. It only affects the behaviour on Windows.
The original fix of years ago must have caused leaked memory, and was
undone by my other changes to track memory management. This fix preserves
the intent, and avoids leaked memory.
Note that CleanDir may be used on a temp directory, deleting entire projects, and in that case
we delete all files too, but do not want to delete files that are at the top level, only ones inside
projects.
Projects will only have symlinks in them if they have been tampered with, and then
following the links, especially when doing CleanDir(), would be dangerous. So don't.
Improved comments, otherwise it isn't clear that CleanDir is needed on Windows
and Linux, not just Mac. It is deleting whole subdirs, not just .DS_Store files.
Fix affects windows only.
In the case of Save and SaveAs creates the directory too, so that the dialog can be positioned there.
For Export Multiple, will prompt if the directory does not exist.
Previously we relied on Windows doing this. With the fix to now use Documents directory, if none was set, Windows no longer remembers the directory. So now we remember it instead.
Added for consistency with log of remaining files. This change should clear up .DS_Store on Mac, and so help with 1567 residual issue when files are not deleted on exit.
On windows the capitalisation does not matter.
On Mac/Windows we use 'audacity' in path names, even if the directories wxWidgets suggests have 'Audacity'.
... At startup, spawn a copy process in main() and crash it at once. The
child process continues. (The fix was ineffective when the parent continued
and the child crashed.)
Re-activiation of the application is also needed or else the windows are
hidden behind other applications.
Another sleep when closing a project, to fix occurrences of the bug caused by
closing of a project window without saving changes.
The short-circuit evaluation of flags ONLY generated the NotMinimized flag, and all others were set to zero, when short-circuit evaluation happened. This change remembers and preserves the other flags, so they don't get set to zero.
Previously this could/would fail to create a temporary directory if the earlier
components in the path did not exist. The user would then be prompted
unnecessarily for a location to store temporary files.
Also now disallow blank as a temporary directory name.
Also no longer search entire dmg for plug ins. Just search the app itself.
I have done as suggested in the bug comments, and used:
Windows: Users\username\Documents
Unix/Mac: ~/Documents .
We could though use GetLocalizedResourcesDir() and could use Music rather than Documents.
Opening Audacity can give a project window that straddles screens (if that is what the user wants and had last time they closed).
New windows created from menu item new will still start out initially all on one screen.
Works by un-full-screening before closing. Patching wxWidgets was needed to
make that work correctly after full-screening by clicking on the green circle
in the title bar, but that fix is not needed for full screen after command+/
... To short selections (under 1/100 second), at the start of the second of
two separated clips, in a project with a high sampling rate of 192000.
Symptom was that the audio moved so it was pasted to the end of the first clip.
Other uses of the constant WAVETRACK_MERGE_POINT_TOLERANCE should be
reviewed too but this is a conservative fix made close to release time.
... Partial reversion of commit 2e8ee5fe75
Now, if you open an effect dialog, Return key will apply it. But, if you
cycle among windows with alt+f6, then the focus moves among the controls in
the effect dialog, and I don't know how to prevent that.
At least so it is in Mac. I have not verified it with Linux.
The MAD decoder will not decode the final frame in an mp3 stream unless
it has a small amount of padding afterwards (MAD_BUFFER_GUARD bytes,
actually 8). Without this, it loses sync before returning any decoded
data from the final frame. The result is that the imported audio is
truncated by up to 1152 samples.
This commit addresses that by using a final round of input callback
calls to provide the last MAD_BUFFER_GUARD bytes after the underlying
file has reached eof. The logic is based on madplay.
