Renaming directories

[Christina Gratorp]

Hi!

I have a question about renaming directories in svn. We work on trunk and 
branch out when we have a release. I just discovered we have a binary with the 
same name as a directory within the same namespace and since renaming the 
binary would indeed make people go mad I must rename the dir. Now my
question is what will happen if I rename the directory on trunk and then later 
on want to rebase from trunk up to an old release branch for bug fixing and 
such. How will svn handle the name change? I've tried reading in manuals but 
unfortunately wasn't able to find anything on this.

Best regards,
Christina Gratorp
Software developer, Transmode Systems AB

Re: Setting auto-props from Server side ?

[Lorenz]

Ravi Roy wrote:
>Just curious if there is a way out to set auto-props for certain binary
>files on server side ? I know, it is can be set on Client like TortoiseSVN.

auto-props are a client side configuration that's shared by all
subversion clients (TSVN only provides a GUI for that task). 

You could put the relevant config file under version control though.
That won't give automatic propagation of changes, but at least you
only have to do an update on the client to get the actual settings (no
manual fiddeling in config files)
-- 

Lorenz

Re: Wanted: pre-commit hook to prevent commits to externals definition

[Lorenz]

Daniel Becroft wrote:
>Lorenz wrote:
>> David Aldrich wrote:
>>>Please can anyone point me to an example of a svn pre-commit hook that 
>>>prevents commits
>>>to files that are members of an externals definition?
>>>
>>>We want to allow such files to be committed only from within the external 
>>>directory, i.e. where
>>>they originate from.
>>
>> As long as the externals are pointing to the same repository it should
>> be possible to examine the transaction and determine if any of the
>> changed paths is related to an external definition (external
>> properties can be extracted from the transaction too).
>
>Really? From my testing, it seemed that SVN doesn't descend into
>externals during a commit, regardless of whether the external points
>to the same or a different repository.
>
>Have you seen different behaviour?

yes and no 8-)

The command line client does not decend into externals, but you can
list the external folders additionally to get an atomic commit,

and I'm using TSVN as my main client, which lists externals from the
same repository in the commit dialog, selected for commit by default.
-- 

Lorenz

Re: Two svn/apache servers accessing one database

[Richard England]

This thread starts to discuss this but it is not clear if anything 
definitive has been done to insure that multiple access is truly 
supported on a single FSFS data base.


http://svn.haxx.se/users/archive-2008-09/0791.shtml

Anyone ?


 Richard England wrote the following on 06/04/2010 04:41 PM:
Are there any possible repercussions of having two server both running 
Apache/SVN (same version)  accessing the same database files?  This is 
using FSFS.


Is this likely to cause data corruption or anything nasty?

~~R

Re: Is there an authenticated no-op in subversion?

[Daniel Danger Bentley]

On Mon, Jun 7, 2010 at 4:27 PM, Rob van Oostrum  wrote:

> On Mon, Jun 7, 2010 at 12:03 PM, Daniel Danger Bentley <
> dtbent...@gmail.com> wrote:
>
>> I'm working on a system to interact with subversion automatically. Some of
>> the repositories in question work over https, and this requires
>> authentication for certain actions.
>>
>> I would love to separate authentication from action. Is there a way to
>> perform an authenticated no-op (which would fail if not authenticated)? If
>> this did exist, then I could first try this, and perform the hairier actions
>> only once I'm sure I'm authenticated.
>>
>> The best we've been able to come up with at work so far is to do a propset
>> on a revision property.
>>
>> Thanks,
>> -Dan Bentley
>>
>
> You are confusing authentication with authorization. Authentication = "user
> is who they say they are". Authorization = "user has permission to do X"
> (user may or may not be authenticated), which can (and often does) depend on
> both authz configuration and repository hooks.
>
> Authentication can be covered simply by disallowing any non-authenticated
> write access. Predicting authorization is more complicated unless you can
> safely "if test X works, real operation Y should also work".
>
> If all you really need is in fact an authenticated no-op, you should be
> able to run any read operation, which - if you are disallowing anonymous
> access entirely - will only work when already authenticated.
>
>
We don't disallow anonymous access entirely. Anonymous read access is often
allowed.

In this case, I control the client software but not the server. So I would
like to test if things work on the client so I can fail early.

So there is no dry run mode I could try an operation through? If only writes
require authorization, then there is no no-visible-impact way to test
authentication?

Thanks much,
-Dan



> Cheers
> Rob
>



-- 
"My youngest daughter is 13 months old, and in her limited vocabulary
alongside essentials like 'mom-mom', 'da-da', and 'meow' she knows to say
'zhooom' and wave her arm around when she wants me to get a lightsaber down
off a shelf for her to play with. As far as I'm concerned that's Parent of
the Year Award material right there." - Aurich Lawson

Re: Is there an authenticated no-op in subversion?

[Rob van Oostrum]

On Mon, Jun 7, 2010 at 12:03 PM, Daniel Danger Bentley
wrote:

> I'm working on a system to interact with subversion automatically. Some of
> the repositories in question work over https, and this requires
> authentication for certain actions.
>
> I would love to separate authentication from action. Is there a way to
> perform an authenticated no-op (which would fail if not authenticated)? If
> this did exist, then I could first try this, and perform the hairier actions
> only once I'm sure I'm authenticated.
>
> The best we've been able to come up with at work so far is to do a propset
> on a revision property.
>
> Thanks,
> -Dan Bentley
>

You are confusing authentication with authorization. Authentication = "user
is who they say they are". Authorization = "user has permission to do X"
(user may or may not be authenticated), which can (and often does) depend on
both authz configuration and repository hooks.

Authentication can be covered simply by disallowing any non-authenticated
write access. Predicting authorization is more complicated unless you can
safely "if test X works, real operation Y should also work".

If all you really need is in fact an authenticated no-op, you should be able
to run any read operation, which - if you are disallowing anonymous access
entirely - will only work when already authenticated.

Cheers
Rob

Re: Setting auto-props from Server side ?

[Hyrum Wright]

On Mon, Jun 7, 2010 at 12:01 AM, Ravi Roy  wrote:

> Hi,
>
> Just curious if there is a way out to set auto-props for certain binary
> files on server side ? I know, it is can be set on Client like TortoiseSVN.
>

This feature (and similar ones) are bandied about as "repository-dictated
config", and are on the roadmap:
http://subversion.apache.org/roadmap.html

There are preliminary design documents here:
http://svn.apache.org/repos/asf/subversion/trunk/notes/repos-dictated-config

There is a current development timeline, as most of the developers are
working hard on getting 1.7 shipped.  Help is always appreciated, though. :)

-Hyrum

Re: Setting auto-props from Server side ?

[kmradke]

Ravi Roy  wrote on 06/07/2010 12:22:52 AM:
> On Mon, Jun 7, 2010 at 10:41 AM, Ryan Schmidt 
 > wrote:
> On Jun 7, 2010, at 00:01, Ravi Roy wrote:
> 
> > Just curious if there is a way out to set auto-props for certain 
> binary files on server side ?

> No, there isn't, not a supported way, anyway.
> 
> This has come up countless times on the mailing list before. I'm 
> sure you can find the discussions in the archives at 
http://svn.haxx.se/users/
> 
> The strategy I recommend is to install a pre-commit hook that 
> rejects commits that do not conform to your guidelines (regarding 
> properties or otherwise). Also install a Subversion config file on 
> each client to automat the setting of properties that match your 
guidelines.
> 
> The other strategy is to write a post-commit hook that sets the 
> properties you require. However this is not recommended because it 
> means the server has to manage a working copy. It also means if a 
> client commits a file that doesn't have the properties, the server 
> will add them, and then the client's file will be immediately out of
> date and the user will have to run "svn up" to get the update from 
> the server. The user won't expect to need to do this so this will be
> confusing to the user.
>  
> Thanks Ryan for quick reply. I would check this throug pre-commit hook.
 
A post-commit example to automatically add the needs-lock property is
available at:

http://svn.apache.org/repos/asf/subversion/trunk/contrib/server-side/add-needs-lock.py

It does not require a server side working copy but does need access to
the python bindings.  As Ryan said, it isn't recommended to have
the server create new revisions because of the potential user confusion
that can result from newly added files becoming instantly out of date.

Assuming your PYTHONPATH environment already includes the path
to the Subversion python bindings, the post-commit on unix would need to:

/path/to/hook/scripts/add-needs-lock.py -r $2 $1


Kevin R.

Re: SVN - Hook Scripts

[vishwajeet singh]

There is no separate documentation as such for hook scripts its part of
Subversion book.
Same can be found here
http://svnbook.red-bean.com/en/1.4/svn-book.html#svn.reposadmin.create.hooks


On Mon, Jun 7, 2010 at 10:29 PM, Venkata Badipatla <
venkata_badipa...@persistent.co.in> wrote:

>  Hi All,
>
>
>
> Please any one of you send me the document of hooks scripts of Subversion.
>
>
>
> *Thanks,*
>
>
>
> *Venkata Badipatla **|** Systems Engineer - PMSG (Professional & Managed
> Services Group) **|** Persistent Systems*
>
> *venkata_badipa...@persistent.co.in* *  **|
> **Cell: +91-9657387160 **| **Tel: **+91 (20) 3023 4085*
>
> *Innovation in software product design, development and delivery - **
> www.persistentsys.com* 
>
>
>
> DISCLAIMER == This e-mail may contain privileged and confidential
> information which is the property of Persistent Systems Ltd. It is intended
> only for the use of the individual or entity to which it is addressed. If
> you are not the intended recipient, you are not authorized to read, retain,
> copy, print, distribute or use this message. If you have received this
> communication in error, please notify the sender and delete all copies of
> this message. Persistent Systems Ltd. does not accept any liability for
> virus infected mails.
>



-- 
Vishwajeet Singh
+91-9657702154 | dextrou...@gmail.com | http://bootstraptoday.com
Twitter: http://twitter.com/vishwajeets | LinkedIn:
http://www.linkedin.com/in/singhvishwajeet

SVN - Hook Scripts

[Venkata Badipatla]

Hi All,

Please any one of you send me the document of hooks scripts of Subversion.

Thanks,

Venkata Badipatla | Systems Engineer - PMSG (Professional & Managed Services 
Group) | Persistent Systems
venkata_badipa...@persistent.co.in  | 
Cell: +91-9657387160 | Tel: +91 (20) 3023 4085
Innovation in software product design, development and delivery - 
www.persistentsys.com


DISCLAIMER
==
This e-mail may contain privileged and confidential information which is the 
property of Persistent Systems Ltd. It is intended only for the use of the 
individual or entity to which it is addressed. If you are not the intended 
recipient, you are not authorized to read, retain, copy, print, distribute or 
use this message. If you have received this communication in error, please 
notify the sender and delete all copies of this message. Persistent Systems 
Ltd. does not accept any liability for virus infected mails.

Is there an authenticated no-op in subversion?

[Daniel Danger Bentley]

I'm working on a system to interact with subversion automatically. Some of
the repositories in question work over https, and this requires
authentication for certain actions.

I would love to separate authentication from action. Is there a way to
perform an authenticated no-op (which would fail if not authenticated)? If
this did exist, then I could first try this, and perform the hairier actions
only once I'm sure I'm authenticated.

The best we've been able to come up with at work so far is to do a propset
on a revision property.

Thanks,
-Dan Bentley

Re: Huge Problem

[David Weintraub]

You don't provide must detail, but I'm going to take a blind stab at it:

Check the ownership and permissions of the files you restored. These
files should be owned by the Subversion server user. If you're using
Apache's http, they should be owned by the Apache user (something like
"apache", "http", or "wwwrun".) If you're using "svnserve", the files
should be owned by the user running svnserve.

Also make sure the files are read/writable by that user. Normally, the
permissions should be 644, but if you're using ssh+svn, they should be
664 before the group needs read/write permission.

On Sun, Jun 6, 2010 at 5:31 AM, Abius X  wrote:
> Hello
> I'm unable to commit or update, since i've restored a previous revision of my 
> subversion repository on the server.
>
> What should I do?
> Regards
> AbiusX
>
>



-- 
David Weintraub
qazw...@gmail.com

Re: Any data compression between server and client?

[Je suis la poubelle]

On Mon, Jun 7, 2010 at 15:02, Hyrum K. Wright
 wrote:
>
> If you are using http or https, there are significant latency issues
> associated with the Subversion protocol, due in part to the number of
> roundtrips made to the server for each connection.   Subversion 1.7
> introduces a new version of the protocol which removes most of the
> latencies, and should speed things up significantly.

 OK, waiting impatiently for that version. :)

>
>>     I was pretty sure there's compression and I'd like to dismiss
>> this as a possible cause.  But when I want to confirm it on Tigris
>> foum, I got a rather unfriendly reply:
>>
>> http://tortoisesvn.tigris.org/ds/viewMessage.do?dsForumId=4061&dsMessageId=2616471
>
> I think the gist of that message was "please let us know where you got such
> information" and "please read the documentation before asking questions
> answered therein".  Both are reasonable requests, though could potentially
> have been framed better.  You are must more likely to get assistance if you
> demonstrate you have done your own research prior to asking questions here.

 If you read the other replies from that gist (or whatever you
call it), he implies that there's no compression in SVN/Subversion.

 I HAD DONE my research prior to ask the question, but it's
impossible to "demonstrate" that fact.  I had searched through a lot
of docs.  If he could plug into my mind, I could show that moron how
much time I had spent on this matter.

Re: Any data compression between server and client?

[Hyrum K. Wright]

On Mon, Jun 7, 2010 at 7:44 AM, Je suis la poubelle wrote:

> Thanks for your reply.
>
> My company has developers aboard accessing our SVN server through
> VPN and they're always complaining that transmissions are very slow.
> That's why I'm trying to find where the problem is.
>

If you are using http or https, there are significant latency issues
associated with the Subversion protocol, due in part to the number of
roundtrips made to the server for each connection.   Subversion 1.7
introduces a new version of the protocol which removes most of the
latencies, and should speed things up significantly.

I was pretty sure there's compression and I'd like to dismiss
> this as a possible cause.  But when I want to confirm it on Tigris
> foum, I got a rather unfriendly reply:
>
> http://tortoisesvn.tigris.org/ds/viewMessage.do?dsForumId=4061&dsMessageId=2616471
>

I think the gist of that message was "please let us know where you got such
information" and "please read the documentation before asking questions
answered therein".  Both are reasonable requests, though could potentially
have been framed better.  You are must more likely to get assistance if you
demonstrate you have done your own research prior to asking questions here.


> Is there any difference between SVN and Subversion, btw?
>

'svn' is the name of the commandline binary, and a common nick-name from
Subversion.

-Hyrum

Re: Any data compression between server and client?

[Je suis la poubelle]

 Thanks for your reply.

 My company has developers aboard accessing our SVN server through
VPN and they're always complaining that transmissions are very slow.
That's why I'm trying to find where the problem is.

 I was pretty sure there's compression and I'd like to dismiss
this as a possible cause.  But when I want to confirm it on Tigris
foum, I got a rather unfriendly reply:
http://tortoisesvn.tigris.org/ds/viewMessage.do?dsForumId=4061&dsMessageId=2616471

 Is there any difference between SVN and Subversion, btw?

 TIA


On Fri, Jun 4, 2010 at 21:12, Mark Phippard  wrote:
>
> Subversion compresses the data it sends over the wire.  Using
> mod_deflate can get you a little extra compression on the entire HTTP
> request.  However, there is a huge memory leak when mod_deflate and
> Subversion are used together and a client that does not support
> deflate is used to access the repository.  So it is best to stay away
> from using mod_deflate and the benefits are relatively small since
> Subversion already uses compression.

Re: Wanted: pre-commit hook to prevent commits to externals definition

[Daniel Becroft]

On Mon, Jun 7, 2010 at 6:52 PM, Lorenz  wrote:
> David Aldrich wrote:
>>Please can anyone point me to an example of a svn pre-commit hook that 
>>prevents commits
>>to files that are members of an externals definition?
>>
>>We want to allow such files to be committed only from within the external 
>>directory, i.e. where
>>they originate from.
>
> As long as the externals are pointing to the same repository it should
> be possible to examine the transaction and determine if any of the
> changed paths is related to an external definition (external
> properties can be extracted from the transaction too).

Really? From my testing, it seemed that SVN doesn't descend into
externals during a commit, regardless of whether the external points
to the same or a different repository.

Have you seen different behaviour?

Cheers,
Daniel B.

Re: Add files mantaining originale file timestamp

[Ryan Schmidt]

On Jun 7, 2010, at 03:53, Andrea Antonio Maleci wrote:

> I would like that add command maintains original file timestamp, instead of 
> assign commit time.
> Is it possible ?

No, sorry, it isn't.

http://subversion.tigris.org/issues/show_bug.cgi?id=1256

Add files mantaining originale file timestamp

[Andrea Antonio Maleci]

Hi,
I'm adding files to a repository.
I would like that add command maintains original file timestamp, instead of 
assign commit time.
Is it possible ?
Thanks,
Andrea

Re: Wanted: pre-commit hook to prevent commits to externals definition

[Lorenz]

David Aldrich wrote:
>Please can anyone point me to an example of a svn pre-commit hook that 
>prevents commits
>to files that are members of an externals definition?
>
>We want to allow such files to be committed only from within the external 
>directory, i.e. where
>they originate from.

As long as the externals are pointing to the same repository it should
be possible to examine the transaction and determine if any of the
changed paths is related to an external definition (external
properties can be extracted from the transaction too).
-- 

Lorenz

Re: Check out problem because of alleged problematic URL

[Ryan Schmidt]

On Jun 7, 2010, at 01:18, Hirschberg, Benyamin wrote:

> 

> Ryan Schmidt wrote:

> 
>> On Jun 6, 2010, at 11:39, Daniel Shahaf wrote:
>> 
>>> Hirschberg, Benyamin wrote on Sun, 6 Jun 2010 at 16:49 -:
>>> 
 [benya...@ada-srp ~]$ svn checkout http://ada-srp/kr/svn/trunk/ kr_repos
 svn: URL 'http://ada-srp/kr/svn/trunk' is malformed or the scheme or host 
 or path is missing
>>> 
>>> That error message comes from neon's ne_uri_parse().  With a client using
>>> serf, or another version of neon, the error might disappear.
>>> 
>>> [[[
>>> /* from libsvn_ra_neon/session.c */
>>> if (ne_uri_parse(url, uri)
>>> || uri->host == NULL || uri->path == NULL || uri->scheme == NULL)
>>>   {
>>> ne_uri_free(uri);
>>> return svn_error_createf(SVN_ERR_RA_ILLEGAL_URL, NULL,
>>>  _("URL '%s' is malformed or the "
>>>"scheme or host or path is missing"), url);
>>>   }
>>> ]]]
>> 
>> Reading that snippet of code, I don't understand why the error would be 
>> triggered. It says it should only occur if the host, path or scheme are 
>> null. In the URL supplied by the user, it seems clear that the host is meant 
>> to be "ada-srp", the path is "/kr/svn/trunk/ kr_repos" and the scheme is 
>> "http:" so I don't see the problem.
>> 
>> What version of neon are you using, Benjamin?
> 
> I'm using "neon 0.28.3". According to the release history it doesn't seem to 
> be particularly old.

Well, it's 21 months old. That *is* kind of old. There have been seven releases 
of neon since then.

However, I'm unable to reproduce the issue on Mac OS X 10.6.3, using Subversion 
1.6.11 and Neon 0.29.3, Subversion 1.6.11 and Neon 0.28.3, or Subversion 1.6.5 
and Neon 0.28.3.

I searched for this error message on Google and found many posts. This one 
seemed possibly relevant:

https://lists.sdsc.edu/pipermail/triton-discuss/2009-July/78.html

It suggests that a second copy of some library (neon? subversion?) somewhere 
else on your system, and referenced by LD_LIBRARY_PATH, could be throwing off 
the version of svn you're actually trying to use. Could that be? If not, search 
Google and see if any of the other hits seem relevant.