Re: Boolean output format - Mailing list pgsql-general
From | Jeff Davis |
---|---|
Subject | Re: Boolean output format |
Date | |
Msg-id | 200210042208.45827.list-pgsql-general@empires.org Whole thread Raw |
In response to | Re: Boolean output format (Garo Hussenjian <garo@xapnet.com>) |
List | pgsql-general |
> I understand. I can implement a wrapper to convert bools, but I'd have to > store the schema in my application. Using php, the database is the only > place where types are actually known! I should check to see if Pear's DB > object does any type-casting at run-time. I suppose you could also have your application consult the catlogs rather than store the schema in the app, but that's annoying as well (extra selects). > > I have thought about this as well. The user-defined types are extremely > useful but they add additional (in this case undesirable) complexity. If we > were talking about anything other than booleans, I would not be making this > case in the first place. True enough. Seems like what you want to do is very basic, so it's not supposed to be hard. > > Is this because of processing overhead or other reasons? I'd argue that the > boolean is an extremely important type, and just as (if not more) common as > date and datetime, and it's certainly less complicated. I was mostly wrong about this one, at least with respect to one developer who also responded. Looks like I spoke too soon. You can read his post, for I don't think now's the time for me to make another interpretation :) > > Now you've hit the root of the problem! I want to be able to forward and > reverse engineer postgres and mysql schemas in php for a basic set of > datatypes (varchar, int, decimal, boolean, date, datetime, and maybe a > couple others). The problem with smallint is that I can translate boolean > to smallint, but it doesn't work in reverse! This is mostly an > interoperability concern. > > I have a database abstraction layer in my application that further > generalizes my queries with some sql abstration... Right now the only > difference between postgres and mysql is the return result of a native > boolean. > I don't entirely understand what you mean, but it looks like you'd like to easily be able to take data from MySQL and put it in PostgreSQL and vice versa. Is there a problem with using smallint? PHP should be able to supply an int to either database or retrieve the result as an int from either database, right? I don't have a MySQL DB readily available, so I'm not 100% sure how they handle booleans. > > I can understand the developers' desire for efficiency in the core. I > should always be weighed against user needs. The answer, in this case, is > unlikely to be a true or false... I'd never expect that the world could be > as precise as our favorite database! See above. I'd say the developers do a good job addressing users concerns, and if they did reject a feature like you describe it would most certainly be for a reason other than efficiency. Again, of course, I don't speak for the developers. Regards, Jeff Davis
pgsql-general by date: