Attention is currently required from: jolly, pespin. falconia has posted comments on this change by falconia. ( https://gerrit.osmocom.org/c/libosmo-netif/+/39280?usp=email )
Change subject: bring twjit into libosmo-netif ...................................................................... Patch Set 5: (3 comments) File include/osmocom/netif/twjit.h: https://gerrit.osmocom.org/c/libosmo-netif/+/39280/comment/dcd4b2b8_4002897c?usp=email : PS2, Line 137: bool osmo_twjit_got_any_input(struct osmo_twjit *twjit); > I'm fine with either of the two proposed. I'll go with the shorter `osmo_twjit_rr_info_valid()` then. File src/twjit.c: https://gerrit.osmocom.org/c/libosmo-netif/+/39280/comment/1b794afb_2f3a7183?usp=email : PS2, Line 504: rtph = osmo_rtp_get_hdr(msg); > I mean dropping ortp completely. I dislike the idea of maintaing both ortp > and twrtp. WOW that is bold! Are you really, truly sure though that twrtp will work as a satisfactory replacement for ortp for all Osmocom users who aren't ThemWi? Considering how I designed twjit for my worldview, and considering evidence from past ML threads suggesting that most people in Osmocom hold views generally opposite to mine regarding how the voice plane should work in a GSM network implementation, I find it hard to believe that non-ThemWi users of OsmoBTS would be happy and content if ortp was suddenly swapped out for twrtp. How do Harald and other senior leaders of Osmocom feel about this issue? Given the magnitude of what is proposed... > Feel free to submit some RFC patch for osmo-bts in gerrit once you want to > receive feedback. 1. I'll be doing this work after the present patches get merged in `libosmo-nettf`, not before; 2. I would need to see full consensus of Osmocom community on the question of doing ortp vs twrtp user-selectable choice, or the super-bold approach of twrtp only. With the OsmoBTS=internal abstraction layer implemented in my current branch patches, it is possible to build without having ortp present, in which case that particular `osmo-bts-*` binary will only use twrtp. But both libraries are still supported at source level. File src/twjit.c: https://gerrit.osmocom.org/c/libosmo-netif/+/39280/comment/6de3a61e_6ec6c4b0?usp=email : PS5, Line 486: msgb_free(new_msg); > I looked into your RTP patch. The received RTP frame are directly sent to the > jitter buffer. I suggest to split it, so the application can decide which > frames are sent to the jitter buffer and which are not. Also it could decide > not to use a jitter buffer at all. (when just doing RTP forwarding) The ability to use twrtp without twjit (non-delayed unbuffered Rx path) is already there: see `osmo_twrtp_set_raw_rx_cb()` in twrtp patch. If you want to do your split idea, you could create a twrtp instance without twjit, then create multiple "standalone" twjit instances, and have your unbuffered/non-delayed Rx callback function do the routing. The only feature you would lose this way, compared to native twrtp with twjit, is generation of RTCP reception reports based on analytics collected by twjit. -- To view, visit https://gerrit.osmocom.org/c/libosmo-netif/+/39280?usp=email To unsubscribe, or for help writing mail filters, visit https://gerrit.osmocom.org/settings?usp=email Gerrit-MessageType: comment Gerrit-Project: libosmo-netif Gerrit-Branch: master Gerrit-Change-Id: Ia3be5834571ca18b68939abbcf1ce3a879156658 Gerrit-Change-Number: 39280 Gerrit-PatchSet: 5 Gerrit-Owner: falconia <fal...@freecalypso.org> Gerrit-Reviewer: Jenkins Builder Gerrit-CC: jolly <andr...@eversberg.eu> Gerrit-CC: pespin <pes...@sysmocom.de> Gerrit-Attention: jolly <andr...@eversberg.eu> Gerrit-Attention: pespin <pes...@sysmocom.de> Gerrit-Comment-Date: Thu, 21 Aug 2025 12:52:24 +0000 Gerrit-HasComments: Yes Gerrit-Has-Labels: No Comment-In-Reply-To: jolly <andr...@eversberg.eu> Comment-In-Reply-To: falconia <fal...@freecalypso.org> Comment-In-Reply-To: pespin <pes...@sysmocom.de>
