My plans for the next weeks

Andrew Bartlett abartlet at
Sun May 22 19:40:24 MDT 2011

I figured I might foreshadow some of the things I hope to work on, when
I get back into a fit mental state to commit patches (Jet lag has been
pretty nasty to me this trip). 

In the short term, I'm keen to work on a few things:

* Sort out the --with parameters in the top level build

The various path-based parameters don't do exactly what they say on the
box, and don't replicate the paths that the autoconf build provides.  I
worked with Andreas on part of this at SambaXP, and now I just need to
read up on what waf does here, and figure out the correct way to
integrate that properly.

* Resolve the last duplicate symbols

This involves dealing with the nt_errstr() and similar functions as
agreed (using talloc_stackframe() in DEBUG()), but I'm still not sure of
the exact approach, as talking with Jelmer after SambaXP, he would
prefer to have less 'magic state' - the talloc_stackframe() itself it
pretty clear, but returning memory on talloc_tos() when that was not a
function argument is not.   

Potentially we may need multiple function wrappers, with one that simply
returns 'unknown error' if the caller truly just wants a constant

The server_id issue is the other blocker here, and then just the
international message display code, which just needs a way to know where
to load the message string files regardless of the codebase used. 

(There will be further duplicates as we bring more code into collision
domains, but we have broken the back of this problem, and know how to
fix it).

* Import the libnetapi and libsmbclient tests into smbtorture4

Once the duplicate symbols are sorted out, then a static build of
smbtorture4 should work, and this is the blocker for this particular

* Implement a s3_loadparm_context() function.

This will contain function pointers to the global, constant result and
parametric options from the source3/ loadparm code.  This could then
allow the lpcfg_*() calls to be handed a valid loadparm context.  This
will allow more code to be made common, while we sort out the loadparm

* Converge set_dyn_*() calls into 'lp_set_cmdline()' calls

Now that we have lp_set_cmdline(), we can override debug levels, log
files and other things without changing the 'compiled in defaults' at
runtime.  This should give us a simpler set of interactions here, and
make it easier to access these settings via a common loadparm

* Deprecation of security parameters in 3.6

I tried (and spectacularly failed) to follow the procedure to get some
deprecation patches into bugzilla for 3.6.  I'll try that again now I'm
past the jet-lag. 

Finally, some of you asked at SambaXP if I was going to 'kick it down
another gear' in terms of my pace of merge patches, as you were worried
you may not be able to keep up with the pace of review.  I agree things
got pretty hectic in the run-up to SambaXP, but while I hope to continue
that work, I won't be pushing out patches at that speed any time soon.  

Andrew Bartlett

Andrew Bartlett                      
Authentication Developer, Samba Team 

More information about the samba-technical mailing list