* Removed duplicate paragraph about instructions for Cygwin.
* Restored Leland's added hyphens for headings. :-)
* Reduced STEP 1 to be less redundant, and to only link to wxWidgets.org, rather than the places they are currently storing downloads, which also reduces their version numbers that we need to maintain.
* Step 2.2: Changed remaining ref to 3.0.2 back to 2.8.12.
* Step 2.3:
** reordered the first two paragraphs, so you get the warning about pht.h before getting instructions that say "Build" (and changed them to be about build configuration, not actually doing the build).
** Added information about "audacity-src\win\wxWidgets_additions\window.cpp" to avoid editing the file.
** Restored the paragraph about the 5 core wxWidgets libs we use. Saves build time, and without that paragraph the subsequent "Because of dependencies..." makes no sense -- the dependencies are specified by those 5.
* Step 4:
** Restored the steps 4.1 - 4.8. Good to have, for those who don't already know how to do it this way.
** Restored the instruction about how to alternatively set WXWIN via cmd.exe, and made it explicit that's what it referred to. For lot of devs, that's easier.
* Step 5.3: Removed refs to LAME and FFmpeg, per discussion on -devel.
* typos and rewordings
I've added some of the new plugin stuff to LV2, Nyquist, and
Vamp so that they play better in the new system. They no
longer get bunched in with the Audacity effects when sorting
or grouping the menus. They have not been fully converted
but they should be good for 2.1.0.
Nyquist plugins now include ";author" and ";copyright"
statements.
Added the 4 new Nyquist plugins to the Windows build.
Audiounits are still coming...had to push them to the back
burner to get this other stuff out of the way.
Scanning for new plugins has been improved so that newly
discovered ones will be shown to the user when Audacity starts.
Effects menu sorting has been fixed and improved.
Disabling effect types in Preferences works again and you
no longer have to restart Audacity for them the change to work.
Effect usage in chains works again.
Plugin registration dialog code simplified a bit.
Group names in the pluginregistry are now base64 encoded. I
never really thought about it, but wxFileConfig group names
are case insensitive and since I was using the group name as
the plugin ID, I ran into a conflict on Linux where there
were two plugins with the same name, just different case. (And
they were different plugins.) Hoping all of this will change
when/if the config file gets converted to XML. (wx3 if finally
including XML support)
A fair amount of cleanup of this new code has been done and
will continue as more stuff is converted.
The big thing is the common efffects UI. Right now Ladspa and VST
have been converted to use it and Audiounits will be next. It makes
everything nice and consistent while reducing the clutter in the
dialog.
Other goodies are:
Ladspa effects now show output controls when supplied by the effect
Ladspa effects now work fine as Analyze type effects
Ladspa now has user presets
VST effects dialog is now less cluttered...leaving more room for the effect
Ladspa and VST effects now share a common UI
Ladspa and VST effects are now usable in chains
Ladspa and VST effects now handle user presets the same way
Currently active effects settings automatically saved and reloaded
Can now do numeric range checking on input fields.
And, as always, plenty of critter squashing.
This change from the old int type to the real enum wxRasterOperationMode was missed during the fix of similar problems in r13403 (when similar fixes were applied), probably because ImageRoll and the affected variable are currently unused and the problem therefore doesn't cause a compilation error and is only fixed preventively.
Compatibility with wxWidgets 2.8 is assured thanks to a conditional definition of wxRasterOperationMode as int in ImageRoll.h, introduced in r13403.
Due to changes in sizing of dialogs with wx3, an assert about the size of the preferences dialog (was max 800x600) fails, which causes an error dialog to pop up. We've recently decided that screens have got bigger, and increased the default size of the Audacity window. Rather than remove this assert, we've upped its limits so that we can go further before we hit the limit. Other code to try to keep the dialog size near to 800x600 has not been changed.
The casting macros wxStaticCast and wxDynamicCast from wxWidgets only work within the hierarchy registered to the wxWidgets class runtime information. But neither the base class wxLog nor the derived class AudacityLogger involved in the cast qualify for this.
This error wasn't visible with wxWidgets 2.8 because asserts weren't activated there by default, which prevented the code that causes the invalid memory access from being executed.
Note: After original creation of this fix, the line causing the crash has been changed in r13426 from using wxStaticCast to wxDynamicCast, which avoids the crash for now thanks to a lucky coincidence. It's still not the desired behaviour (before this fix the runtime type check only checked for wxEvtHandler instead of AudacityLogger) and could start to crash again, depending on implementation details of wxDynamicCast. Hence this fix was still needed, and it was adjusted accordingly.
wxWidgets 3.0 has added a lot of asserts to detect incorrect usage
of its APIs. Now capturing the cursor, when it's already captured,
throws an assertion failure. As the assertion failure window
appears at a very inconvenient moment (when the cursor is captured),
this can lead to Audacity crashing or can even render the whole
Desktop Manager unusable until you find a way to kill
Audacity without using your Desktop Manager (e.g. ssh login)!
The recapturing can occur in the track panel sliders because
the sliders capture the cursor themselves (because they are
also used in other places outside the track panel, where
this is actually necessary), but the track panel also manages
the cursor capturing because it needs it for other operations.
Fixed the recapturing problem by letting the sliders capture the cursor
only if necessary.
The main operational difference is that for v4 effect onward, the
sound name will now be *TRACK*. This will not affect existing
effects since they use version number 3 or less.
This also provides the Nyquist effect with much more information about
the current processing:
Variable Property What
*AUDACITY* VERSION current Audacity version number
*SYSTEM-DIR* BASE Audacity install path
*SYSTEM-DIR* DATA Audacity data path
*SYSTEM-DIR* HELP Audacity help path
*SYSTEM-DIR* TEMP Audacity temp file path
*SYSTEM-DIR* PLUGIN Audacity search path for Nyquist plugins
*PROJECT* RATE current project sample rate
*PROJECT* TRACKS total number of tracks in the project
*PROJECT* WAVETRACKS number of wave tracks in the project
*PROJECT* LABELTRACKS number of label tracks in the project
*PROJECT* MIDITRACKS number of midi tracks in the project
*PROJECT* TIMETRACKS number of time tracks in the project
*SELECTION* START start time of current selection
*SELECTION* END end time of current selection
*SELECTION* TRACKS number of tracks in the current selection
*SELECTION* CHANNELS number of channels in the current selection
*SELECTION* LOW-HZ low frequency from spectrogram (if available, else nil)
*SELECTION* CENTER-HZ center frequence (calculated) (if available, else nil)
*SELECTION* HIGH-HZ high frequence from spectrogram (if available, else nil)
*SELECTION* BANDWIDTH bandwidth in octaves (calculated) (if available, else nil)
*SELECTION* PEAK-LEVEL peak amplitude for the current selection
*TRACK* INDEX 1-based index of track being processed
*TRACK* NAME name of track
*TRACK* TYPE type of track: wave, midi, label, time
*TRACK* VIEW track view: Waveform, Waveform (dB), etc.
*TRACK* CHANNELS number of channels in the track
*TRACK* START-TIME start time of track
*TRACK* END-TIME end time of track
*TRACK* GAIN track gain
*TRACK* PAN track pan
*TRACK* RATE sample rate of track
*TRACK* FORMAT sample format: 16 (int), 24 (int), 32.0 (float)
*TRACK* CLIPS list of start/end times for clips for each channel
