[CM] Freeing up s7 instances & memory

[Iain Duncan]

HI Christos, I'd love to look at the repo if you're able to add me as a
viewer. I'm iainctduncan on github. I look forward to seeing it in action!

On Thu, Sep 10, 2020 at 4:27 AM Christos Vagias <chris.actondev at gmail.com>
wrote:

> Some context about my scenario, where only the proper "free" will work.
>
> I made some small skeleton for making VST3 applications with s7 doing the
> DSP & GUI (using iPlug2
> framework & Dear ImGui for the graphics).
>
> Normal scenario:
> The DAW host loads my dynamic library and whenever a user adds an instance
> of this VST, a new
> s7 instance is created. (well and one more for the GUI, cause it runs in a
> separate thread).
> So we have 2 s7 instances per.. plugin instance.
>
> Now, when a user adds a new instance of  the plugin, 2 more s7 instances
> will be initiated.
> One thing to note here: globals are shared amongst these plugin instances.
> So indeed, a pool of s7 instances could work: when removing a plugin, 2 s7
> instances are freed (the dsp and gui).
> Then, when the user adds a new plugin, these 2 instances of the pool could
> be reused.
> The memory keeps growing though (has to do with the max loaded plugin
> instances)
>
> However:
> It's possible that the DAW runs the plugin sandboxed in a separate process.
> That means globals are not shared, and the s7 instances pool cannot work.
> The memory keeps growing even more quickly in this scenario.
>
> Sorry for insisting a lot on this.
> At the moment even with the worst case scenario of memory growing it's
> fine, cause these VSTs plugins are
> only a playground for me.
> But if this project were to be something used by others, this memory
> leaking is a deal-breaker.
>
> PS
> If anyone is interested in this (VST3 with S7) feel free to contact me (I
> have the repo private still).
> And hopefully they could help with either code or advice :)
>
> On Thu, 10 Sep 2020 at 12:59, <bil at ccrma.stanford.edu> wrote:
>
>> Hmmm -- let me think about that -- I already have a
>> "shadow_rootlet" that masquerades as rootlet (for
>> FFI writers' convenience).
>>
>> _______________________________________________
>> Cmdist mailing list
>> Cmdist at ccrma.stanford.edu
>> https://cm-mail.stanford.edu/mailman/listinfo/cmdist
>>
> _______________________________________________
> Cmdist mailing list
> Cmdist at ccrma.stanford.edu
> https://cm-mail.stanford.edu/mailman/listinfo/cmdist
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://cm-mail.stanford.edu/pipermail/cmdist/attachments/20200910/d7ba56f1/attachment.html>