How does raised cosine pulse shaping work in MATLAB?

I know that Raised Cosine Filter is creating a low-pass Nyquist filter that transforms a stream of discrete bits into analog waveforms.

I'm confused about the parameters in MATLAB's rasied cosine filter. I know that the impulse of the filter needs to be truncated because it's infinite. But I think the filter should have a length of Filter span in symbols? The MathWorks website says the filter has a length of Output samples per symbol * Filter span in symbols +1. I am a bit confused how Output samples per symbol is involved in the length of the filter.

Also, if the filter has a length of L, and the stream of bits have a length of N, then convolution between this two arrays should have a length of N+L-1 right? but I noticed that the output waveform of raised cosine filter has a length of N*Output samples per symbol. Why is that? Does MATLAB generate an impulse response and do convolution?

New to signal modulation, any hint is appreciated!

[Vinayak]

In Raised cosine filtering the length of the output waveform is determined by length of the input symbol and the rate of oversampling of the output symbol i.e. N * Output symbols per symbol.

Moreover, Raised cosine filtering is a "finite response filter" which means that the filter response is subjected only to a fixed length and this length in turn depends upon the filter span and number of output symbols per symbol.