Sending attachments in email using mailer.py script
Let's say I have an exisiting repository at /var/lib/svn/projectA with (tags, branches, trunk structure) on my Linux system. I am using the mailer.py script and mailer.conf file [1] in my post-commit hook for sending out emails whenever a new release tag (like projectA/tags/REL-1.0) is created and that is working fine. My quesiton: Is there a way I can send a file whose location is at projectA/tags/REL-1.0/source.zip as an attachment in the emails that get sent (using mailer.py) whenever a new tag is created? Thanks [1] https://svn.apache.org/repos/asf/subversion/trunk/tools/hook-scripts/mailer/
Re: Sending attachments in email using mailer.py script
I guess my biggest hurdle at this time is how to extract the file from the
repository that I would like to send it as an attachment.
Most of the examples on the net simply specifies an explicit filename (to
show as an example) like this:
*
#!/usr/bin/python
import smtplib
..
filename = "/tmp/test.txt"
.
# Define the attachment section
part3 = """Content-Type: multipart/mixed; name=\"%s\"
Content-Transfer-Encoding:base64
Content-Disposition: attachment; filename=%s
...
%s
--%s--
""" %(filename, filename, encodedcontent, marker)
message = part1 + part2 + part3
try:
smtpObj = smtplib.SMTP('localhost')
smtpObj.sendmail(sender, reciever, message)
print "Successfully sent email"
except Exception:
print "Error: unable to send email"
*
So basically I would want the path
/var/lib/svn/projectA/tags/REL-1.0/source.zip in the filename variable.
Re: Sending attachments in email using mailer.py script
> > You can checkout the file to /tmp and include it from there. You > cannot reference a file in your repository with a file path. I had thought about that but it would mean that I will need to include a svn checkout command to /tmp/ directory in my post-commit hook, right? If that is the case indeed then I am not sure how to get the checkout path of the latest tag commited in my post-commit hook. As an example: So the $REPOS in post-commit has path to the project - for example: /var/lib/svn/projectA. How do I get the path of the tag (so that I can concatenate with $REPOS) that is just being commited - for examle: tags/REL-1.00.
Re: Sending attachments in email using mailer.py script
OK I have managed to get the name of the file (after checking out to the /tmp directory) to be attached in the commit email in a variable called $ATTACHFILE in my post-commit hook. But here is another issue that I am now running into: How do I pass the value of the variable $ATTACHFILE to the mailer.py python script without wrecking the order of the parameters it expects? Thanks
Re: Importing Existing Repository into a New Repository
> > svnadmin dump -rHEAD /path/to/oldrepo > dump > svnadmin load /path/to/newrepo < dump > Thanks Ryan! That worked!
Why doesn't 'svnadmin create' creates tags, branches, trunk?
So I am relatively new to subversion as compared to most of the folks on this mailing list. After seeing lot of examples over the web and reading official subversion book it seems that almost every body creates their repository in some or the other way with tags, branches and trunk structure even though for Subvrsion itself these concepts do not exits. So why don't we have (read as a feature) something like this: # svnadmin create --tbt new_repo and a repositroy with following strcuture is created (automatically) # ls /var/lib/svn/new_repo branches tags trunk Basically the optional arugment "--tbt" in the above command creates the sub-directories tags, branches, trunk automatically for you. It would be a nice (raed useful) feature to have. What do you guys think? Thanks!
Re: Why doesn't 'svnadmin create' creates tags, branches, trunk?
I just noticed that the CollabNet's Subversion Edge has this feature when you create the repository from the web browser interface. I guess at least I am not the only one (perhaps) who thought about this.
Better way to get the full rule of the last commit?
All my repositories live under /var/lib/svn/. Let's the output of the following command (on the SVN server): # svn changed /var/lib/svn/projectA/ is A PartA/tags/DEV-1.00_RC5/ Now in my post-commit hook I need the following value in a variable (say EMAIL_URL): EMAIL_URL=/var/lib/svn/projectA/PartA/tags/DEV-1.00_RC5 Right now the way I am getting this value in my post-commit hook is as follow: DIRCHANGE=`$SVNLOOK changed "$REPOS" | $GREP "^A\W.*" | cut -d' ' -f4` SRC_CO_PATH="$REPOS/$DIRCHANGE" Is there any command which will return the full path of the directory inside that repository under which the changes were made by the last commit?
Re: Better way to get the full rule of the last commit?
> > I assume you mean to also pass "-t $TXN" to svnlook in there somewhere. > Yes. I just copied the command from the shell prompt although I noticed that svnlook command works just fine without the "-t $TXN" argument inside the post-commit hook. > Or are you asking if there's a simpler way? Yes because I think that the above method is not very flexible and scalable and also error prone.
Re: Better way to get the full rule of the last commit?
Daniel, >That is not possible for SVN to determine, as it does not know *how* you are serving the repository (HTTP, HTTPS, svn+ssh, svn, >etc). $REPOS will give you the physical path to the repository, not necessarily the publicly accessible path. That not a problem. I can always prefix the protocol like file:///, http://, https:// or svn:// to the full URL. For example if somehow svnlook can somehow return this: /var/lib/svn/projectA/PartA/tags/DEV-1.00_RC5 say in variable called URL then I do this: URL="file:///$URL"
Permissions to create branches/tags to certain users only
Our repositories lives in /var/lib/svn/ on a Linux server. We use the following sturcture on per project per repository basis: /var/lib/svn/projectA/tags/ /var/lib/svn/projectA/trunk/ /var/lib/svn/projectA/branches/ /var/lib/svn/projectB/tags/ /var/lib/svn/projectB/trunk/ /var/lib/svn/projectB/branches/ and so on... We used LDAP authentication method to authorize users. My question: We need to give permissions to only certain users to be able to create branches and tags so that we can restrict all the developers to the trunk itself. Thanks.
Re: Permissions to create branches/tags to certain users only
David: The standard Python pre-commit hook that comes with Subversion's > source tarball will do the job. Are you referring to the files svnperms.conf.example and svnperms.py found at [1]. > I have a Perl version that does the same thing. > Thanks for sharing your scripts. I will take a detailed look into it. [1] https://svn.apache.org/repos/asf/subversion/trunk/tools/hook-scripts/
Having a path name after specifying repository location
Hi, My repository path is /var/lib/svn for the SVN server. However I encountered a unique situation as follow: The following works: #svnadmin create /var/lib/svn/projectA >From a svn client: #svn co http://svnserver/svn/projectA projectA However the following does NOT work: #mkdir /var/lib/svn/projectB/ #sudo svnadmin create /var/lib/svn/projectB/partA #sudo svnadmin create /var/lib/svn/projectB/partB >From a svn client: #svn co http://svnserver/svn/projectB/partA partB Is it true that all the directories directly under /var/lib/svn/ should be an *actual* repository? Can't we have some pathname like "ProjectB" under /var/lib/svn/ and then have repositories (like PartA and PartB) under it without changing the root path of the repository which is /var/lib/svn? How do you overcome this situation? I want PartA and PartB to be separate SVN repository but I do not want to place them under /var/lib/svn/ directory because they beloing to ProjectB and in future other projects like ProjectD could have PartA and PartB in the same manner. Thanks.
Re: Having a path name after specifying repository location
oops...I had a typo: > However the following does NOT work: > #mkdir /var/lib/svn/projectB/ > #sudo svnadmin create /var/lib/svn/projectB/partA > #sudo svnadmin create /var/lib/svn/projectB/partB > > From a svn client: > #svn co http://svnserver/svn/projectB/partA partB > However the following does NOT work: #mkdir /var/lib/svn/projectB/ #sudo svnadmin create /var/lib/svn/projectB/partA #sudo svnadmin create /var/lib/svn/projectB/partB >From a svn client: #svn co http://svnserver/svn/projectB/partA partA I get the error message: SVN: Could not open the requested filesystem.
Re: Having a path name after specifying repository location
Sorry for bumping this one...We really need to find a solution/workaround for this in order for our SVN implementation to be final. Thanks in advance! On Mon, Oct 18, 2010 at 6:00 PM, Tech Geek wrote: > oops...I had a typo: > > >> However the following does NOT work: >> #mkdir /var/lib/svn/projectB/ >> #sudo svnadmin create /var/lib/svn/projectB/partA >> #sudo svnadmin create /var/lib/svn/projectB/partB >> >> From a svn client: >> #svn co http://svnserver/svn/projectB/partA partB >> > > However the following does NOT work: > #mkdir /var/lib/svn/projectB/ > #sudo svnadmin create /var/lib/svn/projectB/partA > #sudo svnadmin create /var/lib/svn/projectB/partB > > From a svn client: > #svn co http://svnserver/svn/projectB/partA partA > > I get the error message: SVN: Could not open the requested filesystem. > >
Re: Having a path name after specifying repository location
>The other way you could do this, if you insist upon PartA & PartB being separate repositories, is to have a different >block for each project, and specify SVNParentPath as /var/lib/svn/ProjectB . Then you could have PartA & PartB set up as separate >repositories - albeit with a lot more management overhead as you add projects/repositories, and potentially more confusion. You mean something like this? DAV svn SVNParentPath /var/lib/svn/projectA SVNListParentPath On AuthBasicProvider ldap AuthType Basic AuthzLDAPAuthoritative off AuthName "Subversion Repositories" .. .. require valid-user DAV svn SVNParentPath /var/lib/svn/projectB SVNListParentPath On AuthBasicProvider ldap AuthType Basic AuthzLDAPAuthoritative off AuthName "Subversion Repositories" .. .. require valid-user and so on Would be nice if someone already has a script (that excepts the name of the repository) that does the job of entering the block in the apache configuration file. Also why svn does not support nesting paths (logical empty folder) under repository location. I am sure many people might have encountered simialar issues especially if they have existing projects before they started using subversion. Thanks for the help.
Re: Having a path name after specifying repository location
Andy: > As I asked earlier, why must PartA & PartB in a given project be > separate repositories? Why is the more conventional approach not > workable for you? > I understand what you are trying to say and I agree but let's just say that my hands are tied without going into too much details. This is how the stakeholders want it. You know how it is in organizations...
How to get svn-bisect?
We are using svn, version 1.6.12 (r955767) on a Linux based machine (SVN Server) and use TortoiseSVN as clients. However, I do not see command svn-bisect on both of them. How can I get svn-bisect just like there is git-bisect?
Re: How to get svn-bisect?
I would like to point out to guys on this list who may be looking for the svn-bisect tool (just like I was) that it is being included in the next release of Debian [1] and in the backported version of Debian [2]. Also see [3] if you are using a Linux distribution that does not include the svn-bisect by default. [1] http://packages.debian.org/squeeze/subversion-tools [2] http://packages.debian.org/lenny-backports/subversion-tools [3] http://search.cpan.org/~infinoid/App-SVN-Bisect-1.1/bin/svn-bisect
Subversion server and client tools compatibility
We have a SVN server running on a Debian Linux box. Running svn --version returns: svn, version 1.6.12 (r955767) compiled Jul 28 2010, 08:58:12 Users uses TortoiseSVN client to access repositories residing on the SVN server. My question is it OK to use the latest TSVN client "TortoiseSVN-1.6.15.21042-win32-svn-1.6.16.msi" or should be stick to the older version of the TSVN client, TortoiseSVN-1.6.12.20536-win32-svn-1.6.15.msi, that we have been using? Although both the server and the client tools are 1.6.x based but i am not sure if the ".x" makes a big difference or not?
Moving an existing Repository to a newly created repository
All our SVN repositories are stored at /var/lib/svn. Let's say we have an existing repository called "ProjectA" i.e. /var/lib/svn/ProjectA. We created a brand new fresh repository called "Projects" like this /var/lib/svn/Projects. Now we would like to rename our existing repository "ProjectA" to simply "A" and then move it (maintaining all the revisions, prop-changes, etc.) under /var/lib/svn/Projects repository. So before: /var/lib/svn/ /var/lib/svn/ProjectA After: /var/lib/svn/ /var/lib/svn/Projects/A I tried using export/import command and also SVN move command from TSVN client but I am not able to figure out what's is the correct way to do this.
Re: Moving an existing Repository to a newly created repository
> You should be able to use regular old mv on the command line for this. So, svadmin create /var/lib/svn/Projects # Create new repo Projects mv /var/lib/svn/ProjectA A # Rename ProjectA to A and then? how do I move repository "A" under "Projects" repository?
Re: Moving an existing Repository to a newly created repository
We are trying to reorganize some our projects and hence repositories so that it makes some kind of logical sense (naming wise). Our existing repository that we want to move i.e. ProjectA has very few commits (4 to 5) in it and we do not care if the revision numbers change as long as we have commit log history (even with new revision number). In any case, this is what has worked for me: svadmin create /var/lib/svn/Projects # Create new repo Projects svn mkdir /var/lib/svn/Projects/A -m "Creating home for Project A" svnadmin dump /var/lib/svn/ProjectA > ProjectA.dump #Dump ProjectA that we want to move svnadmin load --parent-dir A /var/lib/svn/Projects < ProjectA.dump Thank you all for your help!
Hook to check for a presence of file before committing
Hi, I am just getting started with SVN and I am running the latest version of subversion on a Windows Server. I am looking for a pre-commit.bat hook that will check for presence of a particular file (say project.xml) before commiting to the repository. If the working copy that is about to be committed does not have that file then the commits are rejected else the commits are accepted. I tried looking into svnlook command but not sure how to that command to look for a particular file. I am using TortoiseSVN client. Thanks
Re: Hook to check for a presence of file before committing
>Perhaps if you explain why you want to check for the existence of this file in the working copy we can help you find a different way of >going about it. OK. Let me try to explain this. The code that we are trying to commit is generated by an IDE - a software development tool. The particular file (project.xml) is usually an optional file that is up to the developers/user to generate. However we would like to enforce a policy where all the developers before they commit their changes make sure that the project.xml is also generated and then only a successful commit occurs. Does this help?
Re: Hook to check for a presence of file before committing
>It is possible to verify that a project.xml is one of the files being >committed because you can use "svn changed" at the time of the >transaction to see what files are being added or modified. It would be >possible to fail the transaction if svnlook changed doesn't have a >project.xml file in the list. Yes that's exactly what I was hoping for. Do you happen to know any template/example in perl/python that would give me some hint on how to achieve this? I do not have much experience with either of the language but can easily learn if I look at some examples. And yes I am looking to implement on server side only. Thanks!
Re: Hook to check for a presence of file before committing
>You want to enforce that the developer cannot commit unless they have this project.xml in >their working copy, though it will never be sent to the repository. That's not true. The project.xml (or whatever file that particular file is) should also be a part of commit. I will again repeat my question again since there is so much confusion: Can we have a pre-commit (preferred) or post-commit server hook that can check for the presence of a particular file (say project.xml) when the users send their working copy to the server to commit? If the file is present accept the commit (along with that file) else reject the commit. I would prefer not to implement this on client side. I never thought this would be so complicated.
Re: Hook to check for a presence of file before committing
>Is it a requirement that the file be changed each time (and thus part of a commit), or just that it exists somewhere on the client side >(versioned or unversioned)? The server will only know about it when it changes since the svn client is smart enough to only send >differences. Great question. Let me walk through a scenario. Let's say if somehow we do manage to implement this hook before the development/project begins then after the first revision/commit the project.xml file will also be committed (because we reject any commit if that file is not there). Now let's say the development further goes on and now it is time to commit second time. At this time our hook script again will check for the presence of project.xml file and will also check if the project.xml got changed or not. If it got changed that means the developer did produce a new version of this file and we accept the commit. If the version of the file did not change then we reject it.
Re: Hook to check for a presence of file before committing
Ryan - Thanks for the sample script. As mentioned before the SVN server is on a Windows server. I will convert it to a Windows environment and give it a shot.
Multiple Repositories under one subversion daemon
I am running subversion server on a Windows machine. Right now I am in the process of converting my exisiting projects to start using subversion and my projects are located at different paths/locations. For example: C:\data\project1 C:\project\bluecat C:\software\abc Now it seems that wehn we start svnserver executable it accepts the root path of the repositories like this: svnserve.exe --daemon --root drive:\path\to\repository\root However in my case all my projects are under different locations and unfortunately I cannot move them under one location because lot of other systems are expecting those projects to be at that location. So I would rather change the new system (svnserver) to adopt to the exisiting layout. Is there any way to get around this? Thanks.
Re: Multiple Repositories under one subversion daemon
>I may have totally misunderstood you. Are the above paths you show working copy paths, or are they existing svn repositories? Those are not existing svn repositories yet. Let's just say that those project folders should be at those locations only and I want to create SVN repositories at each of those locations. I am using a Windows XP Pro machine. Also by symlink do you mean Windows shortcuts?
Re: Multiple Repositories under one subversion daemon
If I understand you correctly this would mean that somebody (user/developer) has to checkout the code for each of the projects from the repository location (say C:\svn\data\repositories) to: C:\data\project1 C:\project\bluecat C:\software\abc so that the above locations reflect the latest code?
Two trunks in one repository?
So the concepts of trunks, branches, tags are transparent to SVN. We are in a situation where we might need to have two trunks in one SVN repository. The reason is that we have a family of projects - say ProjectA, ProjectB, ProjectC and so on, each one has it's own repository and have just one trunk (normal setup) since the each of these project has just one part. But now let's say we have a ProjectD which has two sub-systems - PartA and PartB whose code is 95% different. So we are thinking to have two trunks inside the ProjectD repository. We would prefer not to create an individual repository for PartA and PartB because we have decided to categorize each of the repository based on the family of Projects - ProjectA, ProjectB, ProjectC, ProjectD. Just wanted to know to get some thoughts from the experts on this mailing list regarding this setup. Any gotachs I need to watch out for? Thanks!
Re: Two trunks in one repository?
I am thinking something like this: ProjectD ProjectD/PartA/trunk ProjectD/PartA/tags ProjectD/PartA/branches ProjectD/PartB/trunk ProjectD/PartB/tags ProjectD/PartB/branches Beleive me or not in our scenario the code of Part A and Part B never gets merged at any point. The only common part is that at the end of each release of Part A and Part B their output file is concetenated into a one single file which is then programmed on a hardware part by an external tool. So for a scenario like that I would like to keep it very simple. Any feedback or comments regarding the above structure? Thanks!
Re: Two trunks in one repository?
Itamar O, yes you are almost there, I have to say very intelligent guess. I am indeed talking about FPGA development. The only difference is we have a 2 part FPGA on our embedded board. PartA and PartB are both FPGA codes which are developed separately for their respective chips and then later they are concatenated into one single file which is then programmed by the vendor tool on our boards. Concatenation is also not a problem since that is done externally. The only major issue is how do we store (keep in mind SVN principles of trunk, branches, tags) the concatenated file (yes we need to store binary objects too) in my proposed repostiroy structure? I would like this to be as simple as it can be because the developers/users of this system are not very SVN savvy. Thanks! On Thu, Sep 9, 2010 at 1:43 PM, Itamar O wrote: > On Thu, Sep 9, 2010 at 11:23 PM, Tech Geek wrote: > >> I am thinking something like this: >> >> ProjectD >> ProjectD/PartA/trunk >> ProjectD/PartA/tags >> ProjectD/PartA/branches >> ProjectD/PartB/trunk >> ProjectD/PartB/tags >> ProjectD/PartB/branches >> >> Beleive me or not in our scenario the code of Part A and Part B never gets >> merged at any point. The only common part is that at the end of each release >> of Part A and Part B their output file is concetenated into a one single >> file which is then programmed on a hardware part by an external tool. >> >> So for a scenario like that I would like to keep it very simple. Any >> feedback or comments regarding the above structure? >> >> Thanks! >> > > That actually sounds extremely similar to the way I set up repositories for > projects in my organization. > Different projects are unrelated, so they reside in separate repositories, > but every project (which is an embedded-device) has software development > and FPGA development, > so the layout within the project is exactly what you described: > proj/Software/[trunk,branches,tags] > proj/FPGA/[trunk,branches,tags] > > If you case is similar, I recommend using the layout you described, and > considering the following points: > * Add proj/Tools (or something like that) to manage the scripts that > concatenate the final image etc. > * If you also use SVN to store the resulting images (we do, although it is > usually not recommended to keep binary outputs in source control), consider > adding proj/Releases for those. > * If you have code-reuse between projects (we have reusable software > modules and IP cores) - look into svn:externals and decide how to manage it > in your set up. I haven't completed handling this in my set up (still in the > process of migrating from VSS..), but I plan to have a dedicated repository > for "shared assets" and have specific projects define the assets they use > based on absolute svn:externals. >
Re: Two trunks in one repository?
Geoff, I think I am beginning to undestand what you are suggesting. Right now I am in process of implementing this setup. At this point nothing exits - no ProjectD, no PartA and no PartB. So I will try to summarize what I have undestood so far: 1. All our SVN repositories lives under the following location: \\myserver\SVN_Repositories\ProjectA \\myserver\SVN_Repositories\ProjectB \\myserver\SVN_Repositories\ProjectC 2. Now each of the Project directories - Project A, Project B and Project C are SVN repositories of their own (TortoiseSVN->Create Repository here) 3. Now Mr. ProjectD comes along with has two sub-parts - PartA and PartB. So I create a new repository ProjectD (TortoiseSVN->Create Repository here) under this (correct?): \\myserver\SVN_Repositories\ProjectD and then create two directories (with their own tags, branches and trunk) underneath ProjectD (all using SVN commands) like this: \\myserver\SVN_Repositories\ProjectD\PartA\trunk \\myserver\SVN_Repositories\ProjectD\PartA\tags \\myserver\SVN_Repositories\ProjectD\PartA\branches \\myserver\SVN_Repositories\ProjectD\PartB\trunk \\myserver\SVN_Repositories\ProjectD\PartB\tags \\myserver\SVN_Repositories\ProjectD\PartB\branches Note that Part A and Part B are not their individual repositories. They are just directories that live under ProjectD which is a repository. 4. Let's say now PartA and PartB code development begings and at some point a tag is created for each of their release: \\myserver\SVN_Repositories\ProjectD\PartA\tags\REL-1.0 \\myserver\SVN_Repositories\ProjectD\PartB\tags\REL-1.0 Now let's say the output of Part A - PartA.xcf and output of Part B - PartB.xcf are concatenated into a file called PartAB-R1.xcf. My questions: Q1. At this point I would somehow like to store this file (PartAB-R1.xcf) into my SVN repository (ProjectD). What would be an ideal location (logically) to store such a file. Should I be doing any kind of merging from tags (REL1.0) of PartA and PartB to create another node called "Combined-REL-1.0" or something like that. I would like to avoid this if possible. Q2. Also when somebody checks out ProjectD I would like that it pulls the PartA and PartB revision history also so that users cab see the "Revision Graph" of PartA and PartB together using the TortoiseSVN client. I hope I am able to describe the whole scenario to you all.
Re: Two trunks in one repository?
>In SVN a working copy does not include the revision history, which remains on the server (as opposed to Git or Hg). Yes. I am aware of that. > In order to see a revision graph (TortoiseSVN feature), you need to communicate with the repository. Yes. True. >In fact, you don't even need a working copy in order to enter the TortoiseSVN "Repo browser" and view the revision graph. In this case I have to view the "Revision Graph" individually for PartA and PartB. In fact I just tried creating the setup (not using externals so far) that I described in my last post and I do not get the "Revision Graph" option when I right click on "ProjectD" repository under the TortoiseSVN->Repo-Browser. However I do get the "Revision Graph" option when I right-click on PartA and PartB individually. I would like to see the "Revision Graph" of both the PartA and PartB together so that the one can get an overview of the entire project in just one snapshot. Is this possible?
Re: Two trunks in one repository?
For some reasons this message was returned to me by the mailer daemon. So resending it again. >In SVN a working copy does not include the revision history, which remains on the server (as opposed to Git or Hg). Yes. I am aware of that. > In order to see a revision graph (TortoiseSVN feature), you need to communicate with the repository. Yes. True. >In fact, you don't even need a working copy in order to enter the TortoiseSVN "Repo browser" and view the revision graph. In this case I have to view the "Revision Graph" individually for PartA and PartB. In fact I just tried creating the setup (not using externals so far) that I described in my last post and I do not get the "Revision Graph" option when I right click on "ProjectD" repository under the TortoiseSVN->Repo-Browser. However I do get the "Revision Graph" option when I right-click on PartA and PartB individually. I would like to see the "Revision Graph" of both the PartA and PartB together so that the one can get an overview of the entire project in just one snapshot. Is this possible?
Wrong value in db/format when creating from TortoiseSVN
I have the subversion server running on a Debian Linux machine. The repositories are residing on a Windows domain shared network drive on the network which is mapped on the Linux machine as: mount -t cifs //software/svn_repositories /var/lib/svn/ -o username=mynet/techgeek,password=**,uid=www-data,gid=subversion,iocharset=utf8 All the developers/users access the repositories using TortoiseSVN client from Windows machine. The problem is that if the repositories were created from TortoiseSVN then upon accessing the repositories (For example http://svnserver/svnrepos) I get: human-readable errcode="160043": Could not open the requested SVN filesystem However, if I create new repositories from the Linux machine (on which subversion is running), then everything works fine. I noticed when creating repositories from TortoiseSVN the value in db/format is "4" (non-working) and while creating repositories from Linux the value is "3" (working). If I change the vaule from "4" to "3" for the repositories that were created by TortoiseSVN then I do not longer get the above error message. Sounds like there is some incompatibility between the subversion engine (on linux) and the TortoirseSVN. I would like to enable developers/users to create repositories using TortoiseSVN client. Do I have to live with this situation or is there a more elegant workaround rather than changing the value of db/format file manually everytime a new repository is created? Thanks
Re: Wrong value in db/format when creating from TortoiseSVN
>I don't know what the repository version numbers are that correspond to each svn/TSVN version (and you didn't mention which >versions you have installed). Linux SVN Server version 1.5.1 (r32289) TSVN - 1.6.10 (32-bit)
Re: Wrong value in db/format when creating from TortoiseSVN
Daniel, > Or do (the Tortoise equivalent of) 'svnadmin create --pre-1.6-compatible' > Actually I would prefer to do the other way around. Can we tell TSVN to create repositories which are compatible with Subversion Engine 1.5 because we would be usually creating repositories using the TSVN. AFAIU, what you suggested is to be executed on the Linux machine, right? Thanks!
Re: Wrong value in db/format when creating from TortoiseSVN
> > Do not put your repository on a shared drive. Do not use file: >> protocol to access it. It is not designed to work in such way. Read >> the documentation (aka svnbook) and configure a proper server. > > > This should be emphasized more. Although the OP said he wanted the users > to be able to create new repositories, it is really a bad idea to be able to > do this through mapped file access (aside from that being a bad idea in > general for multiuser use) because it also means that anyone can delete or > overwrite your entire history with an inadvertent typo or mouse wiggle. Once the repository is created the users will be accessing it either through svn:// or http:// and NOT with file:/// Although I still want the users to be able to create repository using TSVN because they are not familiar with Linux at all and also that they won't have to wait for some admin person (let's say myself) to just create a repository for them.
Re: Wrong value in db/format when creating from TortoiseSVN
Les, > The point remains that if they have write access through mapped files, >> they have the capability of destroying everything there. >> >> You could probably whip up a cgi script to create new repositories from a >> web request form if something like that doesn't already exist. Or you could >> put more than one project inside a single repository so all you have to do >> is create a new top level directory with the trunk/branches/tags structure >> under it. There are some downsides, but it is not an unreasonable approach >> and it lets users do everything remotely. > > I do undestand your concern but in our case we don't have any option other than putting repositories on the network drives. Let's just say our hands are tied. Having just a top level repository is also not viable in our case. We need one repository per project. My only other option now is to try upgrading the SVN server on the Linux machine...
Re: Wrong value in db/format when creating from TortoiseSVN
OK guys so I ended up upgrading the subversion server to 1.6.12 the same as TortoiseSVN is built against. Now I am able to checkout the repositories through TSVN but whenever I try to commit I get the following error message: $Commit failed (details follow): $Couldn't perform atomic initialization $database is locked This is what I did: 1. I mounted the windows shared network location like this: mount -t cifs //software/svn_repositories /var/lib/svn/ -o username=mynet/techgeek,password=**,uid=www-data,gid=subversion,iocharset=utf8 2. Created a repository: cd /var/lib/svn svnadmin create projectA Edited the svnserve.conf and passwd file to enable authentication for svnserve. 3. svnserve -d -r /var/lib/svn 4. Checked out the repo from TSVN which worked fine 5. On committing from TSVN I get the above error message. 6. I also tried mounting as: mount -t cifs //software/svn_repositories /var/lib/svn/ -o username=mynet/techgeek,password=**,iocharset=utf8,file_mode=0777,dir_mode=0777 but I get the same error message. However if I create the repository on the local Linux filesystem instead of shared network drive, then everything works fine from TSVN. Note: I am using svn:// protocol to checkout and commit. Any ideas?
Re: Wrong value in db/format when creating from TortoiseSVN
I also experimented with http:// protocol and the same error message. I have spent last 2 days just working on these issues and here is what I would like to summarize: 1. Repositories created on shared network drive with 1.6.12 subversion engine (svnadmin) or 1.6.10 TSVN, will not commit through TSVN or even through a Linux svn client. The error message is: svn: Commit failed (details follow): svn: Couldn't perform atomic initialization 2. Repositories created on shared network drive with 1.5 subversion engine (svnadmin) will work fine including committing. So the question is why would repositories on shared network drive will work if they are created with 1.5 subversion engine and will not work if they are created with 1.6? Is this a bug?
Re: Wrong value in db/format when creating from TortoiseSVN
Neil, > Not that I can help, but I wanted to clarify: you have a repo on a > CIFS-mounted shared drive, and are then using a Linux SVN server? > Yes that is correct. Although I admit that may not always be practical, esp. in an enterprise (we > plan to move to using NFS-mapped repos. when our Solaris server's main drive > fills up; some basic trials worked OK, albeit more slowly). > Yes, I have no other option than to put repositories on network drive. From what I have read in the docs it seems that it is not encouraged to use file:/// protocol to access network repositories. I think as long as you use svn:///, http(s):///, etc. protocols to access repos on network share it should be fine. > It could be that some new function or sequence of 1.6 repos fails across > CIFS mounts, maybe something file-locking related. > What would be a good place to report this to SVN folks?
Re: Wrong value in db/format when creating from TortoiseSVN
> > Can you actually use a new repository created by svnadmin 1.5 with your > 1.6.12 server install? Or is it just previous repositories that are working? > > Yes. With my 1.6.12 server install, if I create a new repository like this: #svnadmin create --pre-1.6-compatible myproject then it works fine. But if I create simply like this: #svnadmin create myproject then I get those atomic transaction error message. Any previous repositories that were created with 1.5 (1.5.1 to be exact) engine works fine with 1.6.12 server install. > could you possible try to downgrade? It seems to have to do with using > NFS shares. > My Linux distribution (Debian) right now either offers 1.5.1 or 1.6.12 as binary packages. Getting a 1.6.11 to install will take some additional work to install because of integrated dependencies and all. I can see easily tell whether the repository is 1.5 based or 1.6 based as follow: For 1.5 the contents of projectA/db/format: 3 layout sharded 1000 For 1.6 the contents of projectA/db/format will be: 4 layout sharded 1000 I guess until somebody finds a workaround I will have to live with creating 1.5 compatible repositories from my Linux machine only.
dumping non-continuous ranges using svnadmin dump
I have a repository called "sandbox" whose HEAD is at revision 20. I would like to dump revisions 1 to15 and then from 18 to19. Basically I want to skip revision 16,17 and 20. The command: # svnadmin dump -r 1:15 -r 18:19 sandbox > sanbox-dumpfile gives error message: svnadmin: Multiple revision arguments encountered; try '-r N:M' instead of '-r N -r M' and the command: #svnadmin dump -r 1:15,18:19 sandbox > sanbox-dumpfile gives error message: svnadmin: Syntax error in revision argument '1:15,18:19' How can I do this? Thanks.
Copying hooks automatically upon repository creation
We use one repository per project. All repository lives in /var/lib/svn/. Also we use the hooks post-commit and pre-commit for every repository. Right now we have to manually copy these two hooks whenever a new repository is created. For example: cd /var/lib/svn/ cp /path-to-my-hooks/* projectA/hooks cp /path-to-my-hooks/* projectB/hooks cp /path-to-my-hooks/* projectC/hooks Is there any way to automate this process?
Re: Copying hooks automatically upon repository creation
Thank you all for sharing the information. Really appreciate it.
Re: Copying hooks automatically upon repository creation
Ryan, > sudo -u "$USER" ln -s ../../conf "$REPO" || exit $? > sudo -u "$USER" ln -s ../../hooks "$REPO" || exit $? > I just realized that the above two commands will fail, if the repositories are on a SMB/CIFS mounted share on the Linux system, with the following message: ln: creating symbolic link conf: Operation not supported How do we get around in this situation?
Re: Setting "tags" to read only ?
>$SVNLOOK changed -t $2 $1 | grep "/tags/" && /bin/echo "Cannot commit to tags" 1>&2 && exit 1 I had a question regarding the above command. How do we know what value is going to be passed in $2 i.e. the name of the transaction so that we can test/debug our script. Can we somehow simulate the above command without actually trying to run it during svn commit? Thanks
Re: dumping non-continuous ranges using svnadmin dump
> > svnadmin dump -r 1:15 sandbox > sandbox-dumpfile > svnadmin dump -r 18:19 sandbox > sandbox-dumpfile2 > > Then when it comes time to load, load the first file, then the second Tried but does not work... # svnadmin load sandbox-restore < sandbox-dumpfile The above command restored it up to 15 revision as expected: <<< Started new transaction, based on original revision 15 * editing path : PartA/trunk/source/README.txt ... done. --- Committed revision 15 >>> and then the following command: # svnadmin load sandbox-restore < sandbox-dump2 gave the error message: <<< Started new transaction, based on original revision 18 svnadmin: File already exists: filesystem 'sandbox-restore/db', transaction '18-a', path 'PartA' * adding path : PartA ... The first method also ended up with the same error message. Any more ideas? Thanks.
Importing Existing Repository into a New Repository
I have an exisiting repository at /var/lib/svn/projectA with (tags, branches, trunk structure) whose HEAD is at Revision 25 I just created a new/fresh repository at /var/lib/svn/projectB with (tags, branches, trunk structure) whose HEAD is at Revision 1 (because of importing the tags, branches and trunk structure). Now I would like to import the projectA repository into projectB with increasing HEAD of projectB up to revision 25. svn import commands is only for unversioned files. svn dump and svn load will increase the revision number. How can I do this? Should I just checkout a WC of projectA and then delete the .svn folder from it and then import it into projectB increasing revision of projectB to 2 which is fine? Is there a better way? Thanks.
Re: Importing Existing Repository into a New Repository
> > and load project a as project b and take all the history with you. > Just create a new project b without adding trunk etc. in revision 1. > I do not want all the revision history of projectA into projectB. I just need to start projectB with the latest code (trunk HEAD of) projectA. May be I am missing something very obvious here...
