On Fri, 2010-04-30 at 01:22 -0700, Miroslav Suchý wrote: > On 04/29/2010 11:57 PM, James Hogarth wrote: > > Hi all, > > > > Great to see the 1.0 milestone reached. Friday afternoons in my team > > we assign to 'personal projects' and I've been thinking about trying > > to help with the postgres side of things during this time I have > > 'free'. > > > > The pgsql head in git hasn't had a commit in 9 months... is this even > > valid anymore? > > Well... So what is status of PostgreSQL in Spacewalk? > Yes, no one touched it for 9 months. We had other priorities. > > But anyway, what we have now: > We partially convert our initial schema. > There is some tool called Chameleon: > https://fedorahosted.org/chameleon > which was written by Jeff Ortel specially for Spacewalk. Chameleon can > translate Oracle schema to PostgreSQL schema. But it is far from > perfect. First - it is completely undocumented. Second - it can not > translate everything. Just look at schema/spacewalk/postgres/manual/* > how many things are converted manually. > And maintaing two sets of SQLs schemas is no way. This will not work.
I understand the complexities of maintaining two schemas. Just to throw this idea out there, assuming the postgres backend was good enough, why bother with the oracle one at all? Redhat usually focuses pretty hard at having their open source projects only depend on other open source projects. Or at very least, only optionally depend on the closed ones. The effort could be spent porting to postgres once and drop oracle support instead of trying to constantly convert code written for the closed source oracle's sql to postgres. Kevin > So > we need something that will be able to translate *everything*. If it > will be enhanced Chameleon or we start from scratch (e.g. by writing > rules for good old sed) - that is question... > > And we just focused on initial schema for now. No one looked on SQL > queries in our code. How do we (automatically) translate it to > postgresql? Some queries are separated from code (hbm.xml) some are > still in middle of the code. How we will handle this? What is the best > way? How this all affect speed of code? Then there are some > infrastructural changes which should be probably done prior finishing Pg > support. > > A lot of question. We should decide some design questions, before we > start coding. > > If you really want to help. You can take a look on Chameleon and > investigate if it is worth of continuing in it and write some > documentation for it. > > Or you can start with some simple Spacewalk bugs and first get overview > of Spacewalk internals, before you start with more complex task, which > Pg support is for sure. > > Or you can do some review of Spacewalk packages for Fedora: > https://bugzilla.redhat.com/show_bug.cgi?id=F-Spacewalk > > Or anything else. But jumping directly on Pg support as you first > contribution is probably not good idea. > _______________________________________________ Spacewalk-devel mailing list [email protected] https://www.redhat.com/mailman/listinfo/spacewalk-devel
