Hi Vadim, Neels,
On Sun, Apr 16, 2023 at 07:37:08AM +0700, Vadim Yanitskiy wrote:
> the patch looks good to me. As I said in the IRC, make sure to update the
> 'titan.ProtocolEmulations.SCCP_commit' in osmo-ttcn3-hacks.git, so that
> `make deps` will pull your new commit.
... and first and foremost, make sure that the fix is submitted to upstream.
The process
for that has changed several times, and was not the same for all TITAN
repositories:
Some were using gerrit, and others github. I believe now all are in gitlab, but
I'm not following closely
> Now that you mentioned GitHub, where I have no access to anymore, I am
> wondering if we should host titan.* forks on Osmocom's own Gitea instance
> rather than on a politically affiliated service like this one? If we agree
> on that, I could take care of the migration.
The reason for hosting those repos on github was that upstream TITAN requested
pull-requests to be placed there for some of their repositories. If that is
still
the case, we should cotinue to have our repo where upstream wants PRs.
If upstream has now migrated to somewhere else, then our repo should be from
wherever PRs can be posted.
IMHO, there's no point in having osmocom forks of those TITAN repositories in
another place from where we
cannot contribute upstream.
> IMO, ideally we should try submitting fixes upstream (to [1] in this case),
> so that the TITAN maintainers can be involved in doing code review too.
yes, that is the normal process. The osmocom forks normally only exist
temporarily
until fixes are merged upstreams - or in exceptional cases where we have code
that cannot be accepted upstream.
> They're generally open for contributions and already accepted lots of
> patches from Harald.
Not just by me, there were other contriubtions from osmocom developers, too.
> Worth mentioning that we have titan.ProtocolModules.BSSMAP in Gerrit, but
> not titan.ProtocolEmulations.SCCP. We could add it to Gerrit, but I doubt
> we will see such patches too often, so probably not worth adding.
I'm surprised that any of it is in our gerrit. That probably was an error or
oversight,
and should not be repeated.
> I am fine with sending patches to this ML and reviewing them here. If we go
> the Gitea route, we could use its pull request feature.
I'd say follow whatever upstream process and post a link here, so osmocom
developers
can review it in upstream. Reviewing only at osmocom before contributing
upstream
makes only sense for large/complex changes where the author doesn't feel
confident
sending them upstream straight away - which doesn't appear to be the case here,
where it's a rater trivial/obvious fix?
Regards,
Harald
--
- Harald Welte <[email protected]> https://laforge.gnumonks.org/
============================================================================
"Privacy in residential applications is a desirable marketing option."
(ETSI EN 300 175-7 Ch. A6)
