My colleagues insist on ploughing every bit of output data they can into subversion. I can't see the value in this practice and would much rather use a file system. The key offenders are build and test output.
We end up with folder structures in subversion like:
project repo/tags/build 1.0/binaries
project repo/tags/build 1.0/installation
project repo/tags/build 1.1/binaries
project repo/tags/build 1.1/installation
project test repo/trunk/version 1.0 tests
project test repo/trunk/version 1.0 test results
project test repo/trunk/version 1.1 tests
project test repo/trunk/version 1.1 test results
It gets messy.
From my point of view, source control should be used on files where incremental revisions are made - tags are used to baseline noteworthy revisions, not to archive big binaries.
I'm a big advocate of using a file system here. Chuck the output on a network drive and let the nightly backup keep it safe. Going through subversion seems like more work for no good reason. We are not using any of the features of the system, but I'm told it is best practice everywhere in the absence of a test management suite.
'Best practice' is a big buzzword where I work so the phrase makes me itchy. Unfortunately, it is used to hand-wave away some fairly fruity ideas without presenting the value added.
I feel that my colleagues are over-engineering things and that they are using subversion because they can. My opinion is outnumbered and that suggests to me that I am wrong.
I feel I am missing something. Is this best practice and if so, why?
EDIT The only reason I have been given for this is that it is harder to delete the files. Something I anticipate a backups and access restrictions would help mitigate.