How should we organize svn?
We can have one trunk/branch/tag and under it a bunch of "projects".
We can have multiple top-level things, each having (optionally)
trunk/branch/tag and under those a bunch of projects.
I say optionally, because I see projects that have "sandbox" as one of
the top-level things, and they don't have trunk/branch/tags - I guess
because they didn't set that up. It's probably close to 0 cost - so
maybe we should always set that up in our organization?
Some things in svn: Code for Java framework, documentation (big book),
website, sandbox, (eventually) code for C++ framework etc.
Release mechanisms can have as a part the creation of a "tag" for the
release. So things we release together should be under one
trunk/branch/tag set.
We could put the uima C++ in together with the uima Java - this might
allow synchronized releasing better. Or we could declare these to be
more "decoupled" and independently release-able.
I'm leaning toward this organization:
Top level nodes under svn ... incubator/uima
site
uimaj (for java framework)
uimac (for c++ framework)
sandbox
docs (for general books, tutorials, etc. - docs specific to uimaj or
uimac would go there)
Under each of these except site would be trunk/branches/tags. Site is
always showing just the lastest, so I'm not sure why we would want to
have branches and tags.
Other opinions?
-Marshall
