Re: [PATCH] Make struct tevent_req opaque 2

On Sat, 2009-02-28 at 16:01 -0500, simo wrote:
> This patch makes struct tevent_req opaque to callers by moving it to
> tevent_internal.h and add helpers functions to set async.fn,
> async.private_data an private_status
> 
> This allows the library developer to change or extend tevent_req later
> without breaking API and ABI.
> 
> I suggest the same change should be done to async_req.

Updated patch.

After some discussion with Volker, we decide to shorten the names of
tevent_req_get_private_data() into _tevent_req_callback_data()
and
tevent_req_get_private_state() into _tevent_req_state()

and created tevent_req_callback_data() and tevent_req_state() similar to
tevent_req_create() so that we abort if types do not match.

Simo.

--

-- 
Simo Sorce
Samba Team GPL Compliance Officer <simo <at> samba.org>
Principal Software Engineer at Red Hat, Inc. <simo <at> redhat.com>

Re: [PATCH] tricky bug in idmap_sid_to_uid().

boyang wrote:
> Jeremy Allison wrote:
>   
>> Haven't forgotten about this, just working on the build farm
>> breakage at the moment. I'll get to this....
>>   
>>     
>  Not a problem. Looks like there are more problems like this. In a
> word, some variables are not initialized, but following code relies on
> the values to go to different code paths.. I'll get back to you when I
> work it out.
>   
I'll fix it in a different patch/patches. This patch itself is Ok, I
think.  And this one address a slightly different problem than the
uninitialization one.  Thanks!

Re: [PATCH] Make struct tevent_req opaque 2

On Sat, Feb 28, 2009 at 07:50:45PM -0500, simo wrote:
> Updated patch.
> 
> After some discussion with Volker, we decide to shorten the names of
> tevent_req_get_private_data() into _tevent_req_callback_data()
> and
> tevent_req_get_private_state() into _tevent_req_state()
> 
> and created tevent_req_callback_data() and tevent_req_state() similar to
> tevent_req_create() so that we abort if types do not match.

Apart from the typos in the checkin message, I like that
patch! Metze, what do you think?

Volker

Re: [PATCH] Make struct tevent_req opaque 2

On Sat, Feb 28, 2009 at 07:50:45PM -0500, simo wrote:
> Updated patch.
> 
> After some discussion with Volker, we decide to shorten the names of
> tevent_req_get_private_data() into _tevent_req_callback_data()
> and
> tevent_req_get_private_state() into _tevent_req_state()

Hmmm. Another look revealed that we have an "enum
tevent_req_state" and now also a macro tevent_req_state.
Probably it would be better to rename the macro
tevent_req_state to tevent_req_get_state to avoid confusion,
and despite the fact it's 4 chars longer 

Volker

Suggestion to improve the documentation of changes in WHATSNEW.txt

Hello dear Samba developers and contributors,

As a "porter" of Samba (building packages for Debian and derivatives),
I'm often puzzled by the identificaton of changes in the WHATSNEW.txt
file.

In Debian BTS, we have several bugs reported against the samba package
(http://bugs.debian.org/cgi-bin/pkgreport.cgi?src=samba). Several of
them have been reported "upstream", thus in Bugzilla.

When a new samba version is released, we of course prepare new
packages and one of the challenges we're facing is closing the
relevant bug reports with proper version tagging (ie saying "this bug
was fixed in package version X.Y.Z-n").

Here and there, it's sometimes difficult to find whether a given bug
is really fixed in a new release and properly close it in our BTS. 

As an example, I can cite Bugzilla #5346 or #5996. Both are fixed in
Samba 3.3.1. Both were originally reported in Debian
(http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=454799 and
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=498704,
respectively).

So, indeed, my upcoming upload of the first 3.3.1 package should close
these Debian bugs.

However, WHATSNEW.txt does not mention them. One (#5346) is maybe
considered a "minor" fix...but the other one is a new feature that
would deserve being documented in WHATSNEW.txt. For sure, as bug

Provision pach: make partitions only not erase others ldb

Dear all,

This is a proposal for making provision --partitions-only not wipping 
out some tdb like secrets, idmap and templates.

As far I understand using --partitions-only should only touch sam.ldb in 
order to changes the partitions and/or the modules list and a few other 
options.
Until now the provision --partitions-only wipeout secrets,idmap and 
template before setting up the partitions and then returning.

As it's one of my first patch chances are that it's won't be good 
directly feel free to comment.

Matthieu.

diff --git a/source4/scripting/python/samba/provision.py b/source4/scripting/python/samba/provision.py
index 91cc074..c99b467 100644
--- a/source4/scripting/python/samba/provision.py
+++ b/source4/scripting/python/samba/provision.py
 <at>  <at>  -1004,23 +1004,23  <at>  <at>  def provision(setup_dir, message, session_info,
                         credentials=credentials, lp=lp)
         share_ldb.load_ldif_file_add(setup_path("share.ldif"))

-     
-    message("Setting up secrets.ldb")
-    secrets_ldb = setup_secretsdb(paths.secrets, setup_path, 
-                                  session_info=session_info, 
-                                  credentials=credentials, lp=lp)

build patches

Hi all,

some volunteer here to push this? 

Thanks
Björn

another build fix

one more ...

Re: [PATCH] Make struct tevent_req opaque 3

On Sun, 2009-03-01 at 13:31 +0100, Volker Lendecke wrote:
> On Sat, Feb 28, 2009 at 07:50:45PM -0500, simo wrote:
> > Updated patch.
> > 
> > After some discussion with Volker, we decide to shorten the names of
> > tevent_req_get_private_data() into _tevent_req_callback_data()
> > and
> > tevent_req_get_private_state() into _tevent_req_state()
> 
> Hmmm. Another look revealed that we have an "enum
> tevent_req_state" and now also a macro tevent_req_state.
> Probably it would be better to rename the macro
> tevent_req_state to tevent_req_get_state to avoid confusion,
> and despite the fact it's 4 chars longer 

Ok given that private_state is not really a state but data, and we
already have an enum called tevent_req_state we discussed on IRC to use
tevent_req_data and change the variable in tevent_req from
'private_state' to 'data'.
Attached a patch that implements this.

Simo.

--

-- 
Simo Sorce
Samba Team GPL Compliance Officer <simo <at> samba.org>
Principal Software Engineer at Red Hat, Inc. <simo <at> redhat.com>

Re: build patches

On Sun, 2009-03-01 at 21:16 +0100, Björn JACKE wrote:
> Hi all,
> 
> some volunteer here to push this? 

Isn't there a risk with 
-        <at> -find $(samba4srcdir) -name '*.o' -exec rm -f '{}' \;
+        <at> -find $(samba4srcdir) -name '*.o' | xargs rm -f
that the path is no longer quoted?

You could move to 

 <at> -find $(samba4srcdir) -name '*.o' -print0 | xargs -0 rm -f 

But I'm not sure even that is safe (and portable).

Andrew Bartlett
--

-- 
Andrew Bartlett
http://samba.org/~abartlet/
Authentication Developer, Samba Team           http://samba.org
Samba Developer, Red Hat Inc.