python: libsmb_samba_internal vs. libsmb

[Volker Lendecke]

On Tue, Jan 08, 2019 at 01:04:48PM +0100, Stefan Metzmacher wrote:
> Hi Tim,
> 
> > I wanted to rename the python module because the naming didn't seem
> > consistent with any of the other python bindings Samba has. I think
> > 'libsmb_samba_internal' made sense when it was unused by any of the
> > samba python code (except some test code). But now (well, soon) these
> > Python bindings will be used by the samba-tool code. It looks a bit
> > strange in samba-tool to have:
> > 
> > conn = libsmb_samba_internal.Conn(server)
> > 
> > E.g. my latest patches start to use the new bindings in more places:
> > https://gitlab.com/catalyst-samba/samba/commits/timb-pysmb-latest
> > 
> > But maybe I've misunderstood something. Why do you want to keep the
> > libsmb_samba_internal name?
> 
> The idea was that we make it clear that we don't expose a stable api
> for external consumers. And we're free to change it without
> consequences, as you did with the directory listing.
> 
> samba-tool and other tools are not external consumers and would be
> changed at the same time as we change the api.
> 
> Volker, you invented the name, do you have any comments on this?

I don't really remember that, but the argument that this is an
internal API that has no stability guarantees is pretty convincing. We
need a clear distinction between things we publish to our users and
things that we can freely mess with. On the C side of things we have
the (admittedly limited) libsmbclient and all the internal stuff that
smbclient and winbind use. We can't restrict ourselves to an ABI/API
for some things.

Volker

-- 
SerNet GmbH, Bahnhofsallee 1b, 37081 Göttingen
phone: 0551-370000-0, mailto:kontakt at sernet.de
Gesch.F.: Dr. Johannes Loxen und Reinhild Jung
AG Göttingen: HR-B 2816 - http://www.sernet.de