Hi All,
I'm working on the initial MetaCarta software grant and delivery, and it seems to me that there are a number of potential paths
this can take, so I'd like to solicit input. (I'm also hoping that the mailing list is actually up!)
The grant is to cover at the minimum all LCF java and jsp sources, mod_aa sources, and end-user documentation sources, that are
owned by MetaCarta. One question is, what else should be brought along as well?
Some background may be in order. LCF has components as follows:
- A crawler service/process
- A tomcat web application, for the crawler UI
- A second tomcat web application, serving access tokens for authorization
- An apache2 plugin, which does the authorization itself, and talks to one of
the tomcat web applications
Individual connectors also may have their own services involved. The FileNet and Documentum connectors have such services, for
example. The reasons have to do with either stability (Documentum), or specific dependencies on specific jar versions
(FileNet). For these connectors, we've used RMI and thus created a pair of specific services for each one (an RMI locator
service, and the actual "server" service).
The MetaCarta source tree described debian packages exclusively. This implicitly controls the build dependencies and the
run-time dependencies. Services are described in terms of /etc/init.d scripts. So I guess the overall question is: how much of
this debian infrastructure should we include in the software grant?
My thoughts are that an Apache debianization will need to be reimplemented in any case, because there are probably a number of
MetaCarta-specific assumptions in these packages. So I'd thought that the proper way forward would be the following:
(1) Take the source code
(2) Build standard apache build infrastructure for that source - probably ant for the java, and whatever apache2 is built with
for mod_aa
(3) Build platform-dependent packaging afresh for all the desired apache
targets, based on (2).
However, it occurred to me that there is a fair bit of useful "how to" information built into the current MetaCarta
debianization. So I need help choosing among the following grant options:
(a) Just the sources, ma'am.
(b) Sources plus the original makefile's, for reference.
(c) The complete debian package sources.
Please let me know what you would think would be most appropriate. (Grant Ingersoll has already chimed in with the opinion that
more is better than less at this point.)
A similar question can be asked about tests MetaCarta has that exercise LCF and individual connectors. These tests are written
in python and expect to be run on MetaCarta's appliance, but they could conceivably be useful as an example. Should they be
included in the grant?
Also, FWIW, the grant will describe files as they appear in MetaCarta's repository today. The presumption is that the files
will be moved to their proper location in the Apache world after the initial import into svn.
A final procedural question involves what copyright, if any, should be on the files submitted in the tar. I am definitely going
to be removing the MetaCarta copyrights; do I want to add Apache copyrights before the grant, or not?
Thanks,
Karl
