2012/3/4 Christoph P.U. Kukulies <k...@kukulies.org>: > At the moment I have everything in the filesystem. The test situation is > that about 500 testprograms have to be run > with product release N against product release N-1. The test programs are > duplicated at the moment in all > places. The deployments of product release builds N, N-1, N-2 are also in > the filesystem.
You may use Git with branches for different versions > If I boil it all down from, say 60 GB to 2 GB (as my first attempts resulted > in), I could generate the testbed on the push > of a button out of the database and assemble it in a tree with two branches, > the "new candidate" branch and the "reference" branch, > run the tests. That all could be web app based in the end. Git use own database solution optimized for performance and allows modify all glue scripts in userspace. You may personalize Git for different purposes. > Just a question on "NORMALIZATION": Is normalization being done on a > database by some operation, like > giving a command: "normalize it" or is it done by design of the data > structures being entered? > Christoph "Normalization" is manually process to minimize redundancy and dependency from data structures. See "Database normalization". You may normalize your table to 3 tables and denormalize to 2 tables. Replace attribute "size" with attribute "version". Primary key first table will be reduced to (filename,version), primary key second table to (md5sum). -- Kit _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users