Enable LADSPA plugins on MSVC

[Rossmaxx]

Title says it. I started from #6240 worked on top of it.

note: From my research, I found that some of these plugins have enabled msvc support upstream, so I would suggest to downstream some of these changes. (thanks to ghost (who apparently deleted his acct) for the upstream changes).

Todo:

• Get plugins to build
  • calf
  • caps
  • cmt
  • swh
  • tap

[DomClark]

Do you need the sys/time.h replacements at all? I can't see anything from that header being used anywhere outside some commented-out code in plugins/LadspaEffect/caps/interface.cc.

[Rossmaxx]

Thanks for your review Dom.

Do you need the sys/time.h replacements at all? I can't see anything from that header being used anywhere outside some commented-out code in plugins/LadspaEffect/caps/interface.cc.

Apparently, after removing the entire thing, it's still building fine (had to comment out the #include though).

I also took a look at most of your concerns and fixed most of those. I'll look at the rest later.

Another problem with this pr is that the plugins build but don't load in the selecter menu (does this got anything to do with the __attribute__ changes?)

[DomClark]

had to comment out the #include though

I'd just remove it entirely, along with the commented out seed function. It's not adding any value sitting there, and it'll be in the Git history if we ever need to look back at it.

Another problem with this pr is that the plugins build but don't load in the selecter menu (does this got anything to do with the __attribute__ changes?)

That would be my guess too. Looking at the code in caps_so_init, it appears to be populating an array with information about the plugins inside. If that function doesn't run, the plugin information won't be available.

[Rossmaxx]

For calf and tap, fixes already implemented in the master branches but need to update submodules. I'll go into more detail in #6765

[Rossmaxx]

opened a pr upstream in swh/ladspa#84

update : it's merged

[theGreatWhiteShark]

@Rossmaxx awesome work! Have you thought about bundling those MSVC powered LADSPA plugins separately? Other Linux-based applications targeting Windows would definitely profit as well. (Availability of LADSPA plugins on Windows is one of main recurring requests at Hydrogen).

In addition, your list of plugins is currently missing lsp plugins. It's a more or less recent collection and maybe the best out there yet. It mainly targets LV2 but a lot of those plugins are available as LADSPA as well.

[Rossmaxx]

@Rossmaxx awesome work!

Thanks.

Have you thought about bundling those MSVC powered LADSPA plugins separately?

It's upto the plugin maintainers to choose. I am upstreaming the changes. Swh has merged my changes too. I'll consider later if more requests come.

In addition, your list of plugins is currently missing lsp plugins.

It's not bundled within lmms either. I am targeting only the bundled ones. I might work on it seperately in the future but can't say anything now.

[theGreatWhiteShark]

It's upto the plugin maintainers to choose.

Well, depends on their license. But most - if not all - of them are GPL 2(+) licensed which means you do not have to ask for permission to distribute (or modify) the sources and artifacts.

It's not bundled within lmms either. I am targeting only the bundled ones. I might work on it seperately in the future but can't say anything now.

You do not ship them? Anyway. Maybe I will bundle them at some point when I have more time at hand.

[Rossmaxx]

This pr is now ready.

[tresf]

On MacOS Ventura, I'm getting the following compilation error:

In file included from /Users/owner/lmms/plugins/LadspaEffect/calf/veal/src/organ.cpp:24:
- /Users/owner/lmms/plugins/LadspaEffect/calf/veal/src/calf/organ.h:160:19: error: no matching member function for call to 'lerp_table_lookup_float_mask'
-        return ph.lerp_table_lookup_float_mask(data, ORGAN_BIG_WAVE_SIZE - 1);

[PhysSong]

@tresf I also see this:

<redacted>/lmms/plugins/LadspaEffect/calf/veal/src/calf/fixed_point.h:238:14: note: candidate template ignored: substitution failure [with U = float]: array size is not a constant expression
    inline U lerp_table_lookup_float_mask(U data[(1U<<IntBits)+1], unsigned int mask) const {

I found the faulty change: https://github.com/calf-studio-gear/calf/pull/212/files#diff-a1c8f62ee19cef03296fa3410166e427bd1b151344594675902c2652846d31eaL232-R232
Reverting those changes fixes the error. I'll check if it breaks MSVC.

[PhysSong]

Reverting those changes fixes the error. I'll check if it breaks MSVC.

Seems like it doesn't break 64bit builds, but 32bit builds fail with this error message:

error C2148: total size of array must not exceed 0x7fffffff bytes

[DomClark]

The build fails unless Perl is available because plugins/LadspaEffect/CMakeLists.txt checks the WANT_* variables instead of the LMMS_HAVE_* variables. While that bug wasn't introduced here, it would be good if you could fix it.

[Rossmaxx]

The build fails unless Perl is available because plugins/LadspaEffect/CMakeLists.txt checks the WANT_* variables instead of the LMMS_HAVE_* variables. While that bug wasn't introduced here, it would be good if you could fix it.

Check line 284 and 285 of root cmakelists.

[Rossmaxx]

Did a minor fix to address the missing perl problem.

[tresf]

Is this file supposed to be here? It looks like it may be a local configuration file that accidentally was added. If not, it seems like it will constantly be outdated with paths such as C:/Qt/5.15.2/msvc2019_64, C:/Strawberry/perl.

[Rossmaxx]

@dan-giddins that's a visual studio ide config file. Also, the commits you pushed now look ambiguous.

[tresf]

I see that the submodule for src/3rdparty/mingw-std-threads was updated. Is that intended? Does MSVC use this?

[Rossmaxx]

@tresf that change came after the push by dan. That's not supposed to be there.

[dan-giddins]

Sorry, I did not mean to push commits to this branch! I'm not even sure why I have rights to do this, or how I did it...

[tresf]

I'm not even sure why I have rights to do this, or how I did it...

It's because GitHub -- by default -- allows project members to to push to PR'd branches -- and you're a project member.

Sorry, I did not mean to push commits to this branch!

Stuff happens. :). The more eyes the better! 🍻

[Rossmaxx]

wait what, why did this get closed now?

[tresf]

Bad merge. It's OK, you'll just have to open a new PR once fixed.

[Rossmaxx]

ahh, i messed up the force push