sctplab / sctp-idata

Developing support for draft-ietf-tsvwg-sctp-ndata
BSD 2-Clause "Simplified" License
6 stars 3 forks source link

Dependency of SS selection on I-DATA support #10

Closed tuexen closed 8 years ago

tuexen commented 8 years ago

Describe what happens when an application tries to select an stream scheduler requiring I-DATA support, but this isn't negotiated. EINVAL might the appropriate errno. Describe what happens if the application selects a stream scheduler requiring I-DATA support before the association is established (on a listening socket, for example), and then the association is established without I-DATA support (since the peer doesn't support it).

This issue was brought up by Marcelo on the tsvwg mailing list.

weinrank commented 8 years ago

Shouldn't an SS always support non I-DATA mode?

tuexen commented 8 years ago

It doesn't make sense to use a stream scheduler requiring I-DATA, when I-DATA is not supported, does it?

tuexen commented 8 years ago

... Or have for a SS a defined behaviour when I-DATA is supported and when not. That would also be a possibility. What do you think?

weinrank commented 8 years ago

I meant b) I choose an SS and when I-DATA is negotiated it runs in I-DATA mode - if supported by the SS.

weinrank commented 8 years ago

To be clear: If the SS doesn't support interleaving, I-DATA-chunks should still be used but no message interleaving will happen.

tuexen commented 8 years ago

Fixed in https://github.com/sctplab/sctp-idata/commit/fe5bff65bf541353690b129dbff570b2ae272df2.