Good question. You could still publish the "data source config" workbook but apply permissions on it so only you can edit it, in conjunction with your SharePoint method.
Right. That's what its permissions are now. The issue is, I have coworkers who also need to access it.
you could create a source code control script.
Good question, I'm a newbie and I'd like to generalize the question. I'd like to place all the source information into Subversion, but not things like (for example) extracts, temporary files, logs, ... so it's not adequate to just version the entire Desktop repository. Is there guidance somewhere for exactly what needs to be recoverable in order to regenerate an entire set of projects/workbooks?
I'm currently working on something similar: how to get our server published materials in source control.
i've written a script that downloads everything from the server. I think there are several such examples kicking about somewhere. My script extracts every packaged file so I'm left with just the twb and tds files. They can then be put into source control.
You only need those two file types as far as I know to recover the workbooks. If any extract is lost, it will regenerate.
We do not keep the original file where datasources are created. To amend a published datasources, we just download it, amend and upload again. There is nothing in the original file you need I think.
You could also just add to source control before you extract and upload.
hope that helps!
Thanks for the pointers. For workbooks that originate on Desktop, it looks like all the required info is in the .twb file. The file is XML so works well with source control, except for the section "thumbnail", which I suppose is a utf8-encoded binhex-style binary that'll change with every small change in the view, regardless of whether or not the underlying workbook logic changes.
Couldn't find any .tds files in my local Desktop repo, but the .twb file defined all the bits of the underlying Data Sources. I'll try just versioning these and see what happens.
We've got a few things in process here: Store Tableau workbooks on GitHub
We store the workbooks on Github and, naturally, they have version control. There's also a script that I'll be published here: https://github.com/ps-data/ps-analytics-public
The script above audits all the workbooks/datasources and takes a daily snapshot so you can see:
-what datasources are part of the wb
-what workbooks haven't been viewed in a while
-store old stuff on S3
Let me know if this helps and I'll be happy to chat in detail about the script(s).
The tds file is in your Desktop repository if you click on a datasource and choose add to saved datasources.
The other way to get just the tds file from a published source for example can by found by unzipping the tdsx file. Basically the tdsx is a normal zip file that contains both the tds and the tde file - any unzip program can get that.
In the script that I've written (C# if that is any help), I'm downloading every workbook and datasource from the server and for every published datasource (we package almost everything), I unzip and extract the tds file and ignore everything else. Good point about the thumbnails - I remove them too as they are pointless space takers in terms of version control. That is why I prefer to keep tds files in backup rather than twb files.
Hope that helps
Would you please put the blog you reference somewhere that is free? Pluralsight is a ripoff.
Could you just copy the contents here?
Ah, thanks all. Let me track down the blog post and code.