all the text types are simply a 32bit length and an array of characters. you need to be aware of the encoding, and null is just another character. [dmb>] Yes, I can see that. What I need is a function that will convert to and from whatever the actual encoding happens to be into Unicode, period. I can handle UTF8/16/32 as long as I know which. single and double real/floats are stored in standard x86 IEEE floating point. [dmb>] Apart from the x86/x64 oddity that seems easy enough (I'm using x64). the date and time types are all fixed lengths, mostly a long int or two. [dmb>] Again, I just need a function to convert to and from one single date/time format that covers everything. I use 2xint64 and no TZ. It covers year 9999 down to milliseconds and then some. numeric in postgres is a bit tricky. its a decimal number stored in base 10000 as a series of short integers, and can represent. does your language even have a decimal or BCD or something numeric type? [dmb>] Of course. I have decimal which is also 2xint64: it equates to about 28 digits. For business purposes I think that's enough, but I may go for something bigger in future. Most likely the simplest solution will be to use string as an intermediate. [dmb>] The point is that none of my code is in C, so I can't get Postgres macros or data types and my conversion routines into the same scope. That's why I'm looking for basic routines that support basic C data types as the go between, so I don't have to write too much C code. Regards David M Bennett FACS Andl - A New Database Language - andl.org -- Sent via pgsql-general mailing list (pgsql-general@xxxxxxxxxxxxxx) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-general
