{9} TNTBase News (6 matches)

News about the xSVN/TNTBase System

Created Reporter Summary Ticket
Description
05/27/10 vzholudev 2010-05-27: The first Live TNTBase demo is online. #128

Although TNTBase focuses on the server-side aspects, just for the sake of demoying a bit, the simple TNTBase web-client is put online. One can find it under the path /Try.html

For example:  http://alpha.tntbase.mathweb.org:8080/tntbase/ores/Try.html

One can do the following: 1. Experiment with XQueries 1. Browse the content of a TNTBase FS 1. Use examples on the right side and useful links. The list of examples will be expanded soon. It's just a very preliminary version.

Note: when there are too many results, the component for rendering and highlighting results may crash the page. So beware.

03/10/10 vzholudev 2010-03-10: TNTBase ver. 0.1 is finally released #117

TNTBase 0.1 is released.

Any comments and suggestions are welcome

03/10/10 vzholudev 2010-03-10: tntbase.org domain #116

Now there is a tntbase.org domain. www.tntbase.org or tntbase.org is the homepage of TNTBase

 https://svn.tntbase.org/repos/tntbase is a location of TNTBase repository

10/12/09 vzholudev 2009-10-12: TNTBase debian package released! #77

Now there is a possibility to download a debian package for i386 architecture. Package has been tested on Debian etch/lenny.

Carefully follow the instructions on the main page in order to install TNTBase in a proper way.

09/28/09 vzholudev 2009-09-28: TNTBase has switched to a newer version of Berkeley DB XML #74

TNTBase now is based on a newer version of DB XML, namely 2.5. The switch has three main benefits: 1) Now Result Sets in DB XML can be copied and used outside of a transaction scope. That means that now one doesn't have to convert results into strings or streams, but can use the in-memory XML representation. 2) DB XML supports automated indexing, that is if XML files have the similar structure, the necessary indices will be added automatically upon commit. 3) Now there is a possibility to use Java (as well as C++ or Python) functions inside XQuery. That gives us the possibility to e.g. query an RDF storage inside XQuery if RDF is more suitable for certain needs.

06/28/09 kohlhase 2009-06-28: TNTBase supports Copy & Branching #58

Now there is consistent support of the copy command in SVN, and as a consequence support of the move command and branches, tags and all this stuff which implies using the copy command.

Let me shortly introduce the problem which persisted before. The point is that when one invokes copy command, usual SVN doesn't copy fulltext of the copy target inside a repository. Instead it creates links to the existing fulltext under source path (from where some files or folders have been copied). And this remains true until the development process gets bifurcated. Whereas for normal usage of SVN it was totally fine, the problem arose when somebody wanted to access the latest revision via TNTBase interface. Since DB XML is used in xSVN for storing only the latest revision, a user could see only source of copied files and folders, but not the target.

Now I've implemented the different behaviour of SVN file system. It copies the fulltext of all XML files that are subject of the copy operation. Thus the user will see the new branch via TNTBase interface on the spot. Sounds easy, but in fact I had to modify a lot, since such behaviour contradicts the SVN philosophy (that is "cheap copies"). For example, I also had to elaborate the mechanism which prevents copied fulltext from redeltifying against original source which was done automatically by SVN, otherwise redeltifying would not make much sense.

Yes, there is an overhead in my approach -- the space of a repository. If we create a lot of branches with multitude of files and do modify only a few of them, then the size of an xSVN repository will differ considerably from the same SVN repository (by "the same" I mean the repositories with the same history and content). But nowadays we do not care too much about space. We care much more about features and performance. Now we can query all our branches at the same time. In case of usual SVN we would have to checkout the whole working copy (thus having repeatable files anyway).

Also let's assume that once we created a new branch, the development went only in this branch. In usual SVN in order to retrieve another branch, it would have to apply multiple reverse deltas against the new branch. In my case this retrieval would be straightforward -- all xSVN would have to do is to retrieve the fulltext from DB XML without any application of deltas.

Note: See TracReports for help on using and creating reports.