Re: Subversion 1.4.4 tarballs up for testing/signing (again)

On Thu, 31 May 2007, David Anderson wrote:

> Hi all,
> 
> Following the last 1.4.4 tarballs are ready for testing.
> The magic revision is r25188.
> 
> http://natulte.net/pub/svn/1.4.4/
> 
> As usual, these packages are NOT yet blessed, distro packagers, please
> don't republish them before they are officially announced as released.
> 
> Signatures from full committers back to the list with coverage info
> (which platforms, backends, ra methods, summary of dependency
> versions...).

Tested:
  [ra_local, ra_svn, ra_dav] x [fsfs, bdb]
  JavaHL, SWIG/Python, SWIG/Ruby, and SWIG/Perl bindings

Platform:
  Fedora Core 6, on Intel x86

Results:
  All tests passed.

Signatures:

subversion-1.4.4.tar.bz2
-----BEGIN PGP SIGNATURE-----

Re: [BUG] svnserve aborts on 'svn log' from 1.4.3 client

On 5/31/07, Hyrum K. Wright <hyrum_wright <at> mail.utexas.edu> wrote:
[snip]
>
> r25242 should do the trick, but I don't have a 1.4.3 server to test it
> against.  DJ, could you test it for me?

Sure, I'll try it tonight or tomorrow (the scenario was hitting a
trunk server with a 1.4.3 client, though).

DJ

Re: SASL for ra_svn on Windows

On Sat, 26 May 2007, Vlad Georgescu wrote:

> Stefan Küng wrote:
> > D.J. Heap wrote:
> >> On 5/26/07, Stefan Küng <tortoisesvn <at> gmail.com> wrote:
> >> [snip]
> >>> That would help, yes. But still not solve the problem: sasl would still
> >>> search the plugins in all the registered paths.
> >>
> >> Hmm, you mean the set_path call just adds that path to the front of
> >> the search list -- it would still search the default install for
> >> plug-ins?
> > 
> > I haven't checked the whole source of SASL, but from what I've seen I'd 
> > have to say yes. I'm not even sure if set_path will add the path to the 
> > front of the list - it could even add the path to the back of that list.
> > I think I'll go with my patch instead, that way I'm sure that the other 
> > paths are ignored.
> > 
> 
> Looking at the Cyrus SASL 2.1.22 sources (specifically the lib/common.c
> file), once you call sasl_set_path() it changes the default path
> callback from _sasl_getpath() (which looks into the registry) to
> _sasl_getpath_simple(), which will simply return the value you provided.
>  So it shouldn't look into any other paths than the one you explicitly set.

Good news!  This sounds pretty safe -- static linkage doesn't seem
necessary, so long as no one is overwriting your Subversion-specific
SASL DLLs.

[PATCH] Rename "--merge-sensitive" to "--use-merge-history" in alldocuments

Hello,

This is a follow-up patch for r25247 to completely rename the option
"merge-sensitive" to "--use-merge-history" in all referring documents.

[[
Rename "--merge-sensitive" to "--use-merge-history" in all referring
documents.

* www/svn_1.5_releasenotes.html
* www/merge-tracking/func-spec.html
  Rename "--merge-sensitive" to "--use-merge-history".

Patch by: bhuvan
]]
--

-- 
Regards,
Bhuvaneswaran

Re: Build Error when compiling python binding on Windows with VC6 at r25159 or later

Hi David,

On 5/30/07, David James <james <at> cs.toronto.edu> wrote:
> On 5/28/07, Joe Swatosh <joe.swatosh <at> gmail.com> wrote:
> > On 5/27/07, Masaru Tsuchiyama <m.tmatma <at> gmail.com> wrote:
> > >

> > apr_uint64_t is now used in the Subversion headers.  The patch below fixes the
> > build errors for me, but the tests don't pass.
>
> Hi Joe,
>
> In the Ruby bindings, I see that you used "%clear long long" to clear
> out the old SWIG typemaps. Does this work in the Python bindings? If
> so, perhaps we should get rid of this hack to python.swg altogether
> and just use %clear long long and/or %clear unsigned long long.
>
> Cheers,
>

I don't see why not.  But I think it probably belongs in apr.swg
instead of svn_global.swg.

For that matter, the stuff I did in svn_global.swg for the Ruby
bindings probably belongs in apr.swg too.  Another change I'd like to
see is to define the typemaps in terms of the apr types instead of
long long and unsigned long long.  I experimented with this, but I was
never able to make it work quite right.

Ultimately, I'd like us to generate a little more synergy among the

Re: [BUG] svnserve aborts on 'svn log' from 1.4.3 client

On 5/31/07, D.J. Heap <djheap <at> gmail.com> wrote:
> On 5/31/07, Hyrum K. Wright <hyrum_wright <at> mail.utexas.edu> wrote:
> [snip]
> >
> > r25242 should do the trick, but I don't have a 1.4.3 server to test it
> > against.  DJ, could you test it for me?
>
>
> Sure, I'll try it tonight or tomorrow (the scenario was hitting a
> trunk server with a 1.4.3 client, though).

Well svnserve doesn't abort anymore, but it doesn't work right -- the
1.4.3 client reports that the socket was closed unexpectedly.

DJ

Re: [PATCH] Rename "--merge-sensitive" to "--use-merge-history" in alldocuments

On Fri, 01 Jun 2007, Bhuvaneswaran Arumugam wrote:

> Hello,
> 
> This is a follow-up patch for r25247 to completely rename the option
> "merge-sensitive" to "--use-merge-history" in all referring documents.
> 
> [[
> Rename "--merge-sensitive" to "--use-merge-history" in all referring
> documents.
> 
> * www/svn_1.5_releasenotes.html
> * www/merge-tracking/func-spec.html
>   Rename "--merge-sensitive" to "--use-merge-history".
> 
> Patch by: bhuvan
> ]]

Thanks Bhuvan, looks good.  +1, please commit.

- Dan

need a libneon.lib for windows xp.

Re: [PATCH] Rename "--merge-sensitive" to "--use-merge-history" inalldocuments

On Thu, 2007-05-31 at 19:04 -0700, Daniel Rall wrote:
> On Fri, 01 Jun 2007, Bhuvaneswaran Arumugam wrote:
> 
> > Hello,
> > 
> > This is a follow-up patch for r25247 to completely rename the option
> > "merge-sensitive" to "--use-merge-history" in all referring documents.
> > 
> > [[
> > Rename "--merge-sensitive" to "--use-merge-history" in all referring
> > documents.
> > 
> > * www/svn_1.5_releasenotes.html
> > * www/merge-tracking/func-spec.html
> >   Rename "--merge-sensitive" to "--use-merge-history".
> > 
> > Patch by: bhuvan
> > ]]
> 
> Thanks Bhuvan, looks good.  +1, please commit.

Committed the patch in r25248.
--

-- 
Regards,
Bhuvaneswaran

Re: Subversion 1.4.4 tarballs up for testing/signing (again)

> -----Original Message-----
> From: David Anderson [mailto:dave <at> natulte.net] 
> Sent: Thursday, May 31, 2007 11:58 AM
> To: SVN Dev
> Subject: Subversion 1.4.4 tarballs up for testing/signing (again)
> 
> Hi all,
> 
> Following the last 1.4.4 tarballs are ready for testing.
> The magic revision is r25188.
> 
> http://natulte.net/pub/svn/1.4.4/
> 
> As usual, these packages are NOT yet blessed, distro 
> packagers, please don't republish them before they are 
> officially announced as released.
> 
> Signatures from full committers back to the list with 
> coverage info (which platforms, backends, ra methods, summary 
> of dependency versions...).

VERIFIED: 
-------
Signature and MD5 Hash of subversion-1.4.4.zip.

Excepting svn_version.h, contents of
http://www.natulte.net/pub/svn/1.4.4/zombieslam/subversion-1.4.4.zip
match http://svn.collab.net/repos/svn/branches/1.4.x <at> 25188 

PLATFORM: 
---------
Windows XP Pro Version 2002 SP2
Intel P4
MS Visual C++ 6.0 SP6 

DEPENDENCIES (Didn't use subversion-deps-1.4.4.zip): 
-------------
apr, apr-iconv, apr-util APR 0.9.12
neon: 0.25.5
zlib: 1.2.3
OpenSSL: 0.9.8d
Apache: 2.0.58
BDB:
http://subversion.tigris.org/files/documents/15/32472/db-4.4.20-win32.zi
p 

TESTED: 
-------
[ fsfs | bdb ] x [ file | svn | http (neon) ]
[ JavaHL ] 

RESULTS: 
--------
All tests pass. 

SIGNATURE: 
-----------
User: Paul T. Burba (pburba)
      <contemplating <at> adelphia.net>
      1024-bit DSA key,
      ID 53FCDC55,
      created 2006/06/21

http://www.natulte.net/pub/svn/1.4.4/zombieslam/subversion-1.4.4.zip:

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (MingW32)
iD8DBQBGXvS/2RaJMFP83FURAhkgAJ45Zad+m2UxbbktbmdbWjK5f8IyqQCfXeMa
vecyXvhbNrjltBzOQysfrGw=
=aJ0x
-----END PGP SIGNATURE-----
Paul B.