MMA as a filter

[sciurius]

From the docs, section 2.3:

MMA is designed to read and write files; it is not a filter.

Writing MIDI data to standard output is not practical, but it would be nice if MMA could read its input from standard input. I have some tools that generate MMA input data and it would be handy to pass the data directly to MMA instead of having to write temporary files.

Suggestion: If no input file is present, read standard input. -f option is required in this case.
Alternative: Allow standard input represented by a lone dash. -f option is required in this case.

Priority: Low.

[bvdp]

I have added this to the TODO list.

It's not quite as simple as one might think .... or perhaps it is. A matter of bypassing a number sanity tests and, probably, adding new ones. I had a quick look and it's probably not to big a deal. I'll have look when I have a bit more time.

Seriously, it is a good suggestion.

[bvdp]

Took a few hours to do this. Seems to work fine.

 - can a few you test the newest developer version an verify

 - if you have some cool scripts which create a mma input on-the-fly and send it to mma via a pipe, could you post them here for others to marvel at

Best to all.

[sciurius]

As far as I can see the standard input functionality works fine. Good job!

A minor point for improvement is the error message:

Code:

mma - < q.mma
ERROR:     No filename given, assuming STDIN. Use -f to set an output filename.

Since you need to explicitly specify STDIN using the single dash, I'd suggest "ERROR: Input from STDIN specified. Use -f to set an output filename.".

[bvdp]

Code:

mma - < q.mma
ERROR:     No filename given, assuming STDIN. Use -f to set an output filename.

Since you need to explicitly specify STDIN using the single dash, I'd suggest "ERROR: Input from STDIN specified. Use -f to set an output filename.".

Ahh, yes ... done. I'd written this on my first coding pass to use "no filename == stdin" but I didn't like it. Too easy for a user to forget the filename and get an "unusual" error message. And since the majority of users will never use this option, the explicit "-" seems to make more sense.

Thanks.

[bvdp]

I've found a minor bug in the latest caused by stdin code. It shows up with the library update options. I have it fixed at my end and will put up a new version (devl) in a few days.