[C++] GetEnvelopeStateChunk() - ensure NeedBig is big enough? (SOLVED)

[nofish]

edit:
Solved with new API addition added in Reaper v6.68:

+ API: expose realloc_cmd_register_buf/realloc_cmd_clear to C/C++ API

Usage info here.

=== original post ===

How big of a buffer / c-string should I pass to GetEnvelopeStateChunk() to ensure to always get the complete chunk and not truncated - as envelope state chunks can theoretically grow 'unlimited' in size with lots and lots of envelope points?

[nofish]

bump, anyone?

Actually this question rises for me whenever wanting to get chunks which can grow relatively big, e.g. also GetTrackStateChunk() (containing all the FX subchunks).

Is there a way to query the current size of the chunk before passing a c-string perhaps, or how do I do this correctly?

[Xenakios]

Taking the API at face value, the only way I can see is to start with some size, check if the call succeeded and if not, try again until it succeeds.

So, something like this, assuming GetEnvelopeStateChunk returns false because the buffer wasn't large enough :

std::vector<char> buf(1024); // initial size 1024 chars
while (GetEnvelopeStateChunk(env, buf.data(), buf.size(), true)==false)
{
  buf.resize(buf.size()*1.5); // grow buffer by 1.5x
}

This would blow up, however, if the GetEnvelopeStateChunk call fails because of some other reason...To handle that, could check inside the while loop if the buffer is being resized to some ridiculously large size.

To make it more efficient, the buffer could be a class member variable or global/static, so it can be reused with the already enlarged size if the operation needs to be done multiple times.

[Meo-Ada Mespotine]

Basically you could try checking if < and > at the beginning of a line match in numbers.
< open a new (sub)-chunk and > close them again.

If the count mismatch, try again.

Counting the envelopes(in Track/ItemStateChunks) or the number of envelope-points(in EnvStateChunks) could also help. If the number of envelopes/envelopepoints mismatch, you can estimate, how bigger the NeedBig should be.

Or featurerequest a "GetSizeOfStateChunk(object)"-function, which gets the size of a statechunk for a track/item/envelope-object in question, which would be the best IMHO.

[nofish]

@Xen

I can do e.g.

char envState[10]; // 100% not big enough
bool gotChunkOk = GetEnvelopeStateChunk(envelope, envState, sizeof(envState), false);

though GetEnvelopeStateChunk() returns true nonetheless.


So your approach doesn't seem to work?

edit:
Sorry, missed this:

assuming GetEnvelopeStateChunk returns false because the buffer wasn't large enough :

[Xenakios]

Originally Posted by nofish

So your approach doesn't seem to work?

Then I guess need to do it like mespotine suggested, check that the buffer contains what is expected. Seems like a really badly designed API...

[nofish]

Originally Posted by mespotine

Or featurerequest a "GetSizeOfStateChunk(object)"-function, which gets the size of a statechunk for a track/item/envelope-object in question, which would be the best IMHO.

I think I'll do that if no other reasonable way comes up. Parsing the chunk everytime to check for correctness seems rather crutchy to me.
Thanks.

[Meo-Ada Mespotine]

I made a feature-request, as I think, it's helpful for SWS and other extensions as well:

https://forum.cockos.com/showthread....45#post2142245

[nofish]

Another thought:

Recently realloc_cmd_ptr was added to the API to be able to grow buffers passed from ReaScript, but we can't use this for internal C++ API calls somehow, or can we?

@mespotine
Thanks for opening the FR.