Forum OpenACS Development: Re: Oracle and PG versions getting out of sync

Posted by Janine Ohmer on
I don't expect someone to get all the nuances of Oracle syntax correct if they aren't working on Oracle. But I'm dealing with things like new columns having been added to the data model for Postgres but not having been added to any of the Oracle files even where it would have been straightforward to do so. If the PG-only folks can do that, and perhaps indicate via comments where queries need changes that they don't know how to make, then those of us still using Oracle would have an easier time completing the work.

I also do think it would be wise to have maturity levels for each version, to at least warn people about the situation. I was foolish enough to see that there were *-oracle.xql files for this package and tell my client that it was already ported, so now I'm going to have to go back to him and explain why it's taking me days instead of hours to get this installed for him.

This is not the first time it has been pointed out that very few are using Oracle these days. That, combined with the fact that part of my current slog is due to someone having copied the *-postgresql.xql files to *-oracle.xql and then just leaving any queries that worked for both in place instead of moving them to a common file, has me thinking about the whole way we are handling this. Given that very few are using Oracle anymore, and that it's unlikely that the community is ever going to bother supporting a third database, is it really necessary to pull all queries out into xql files? It is looking to me like it would improve usability for all, especially PG people, if the PG queries were in the Tcl scripts like they used to be and only Oracle-specific queries were put into xql files.

This would be a huge departure and reversal of what we have been doing, but I think in the long run it would be an improvement. Not that anyone wants to go through the entire code base and move queries around; it would be something that people would do a bit at a time as they were working in that area anyway.

Just a thought, not even a suggestion at this point - was curious if anyone else thinks this might be a good idea.

Posted by Andrew Piskorski on
So, the real problem you are having Janine, is that some OpenACS developer was sloppy, and knowingly so. Maybe he had a perfectly good excuse for that, maybe not, but in either case I think it entirely appropriate to point to the specific code in question. Problems not specifically identified are problems that don't get solved.