On Fri, Jan 23, 2004 at 06:23:25AM -0800, Randal L. Schwartz wrote: In that case, I shall. Hmm. 1Mb isn't a lot in my books. But I won't be using SQLite anyway. I've been trying to tell people this for years. And then SQLite came along, and suddenly MySQL looks feature-rich by comparison. There's obviously a market for lightweight vaguely-indexed data-store libraries, but why do they always start trying to go above their station and become a fully-fledged database? Pick a niche market and stick to it, dammit! Grr.. I'm just waiting for SQLEvenLiter to appear once SQLite decides to add things like concurrency and locking. I hear MySQL has finally got around to supporting transactions and finer-grained locking, so evidently it's time for SQLite to move up and fill its place. tjc. -- Turning and turning in the widening gyre The falcon cannot hear the falconer; Things fall apart, the centre cannot hold; Mere anarchy is loosed upon the world.

Is that with a BYTEA field or a TEXT field? TEXT fields can't handle character 0 but BYTEA fields should. In any event what I personally do is Base64 encode before sticking such data in the DB. It's a little more space, but it avoids quoting issues. Matt

On Sat, Jan 24, 2004 at 11:17:40AM -0000, Matt Clark wrote: That's a BYTEA field, yes. Looks like the DBD::Pg module doesn't handle the escaping properly or something. I'm happy to stick with BLOBs, partially because they avoid the memory overhead of having a 20Mb file stored in memory as a return from the DB. Blobs give you a file descriptor, so i can just iterate until eof, reading and printing to the client connection. tjc -- Turning and turning in the widening gyre The falcon cannot hear the falconer; Things fall apart, the centre cannot hold; Mere anarchy is loosed upon the world.

On Fri, Jan 23, 2004 at 02:58:02PM +0000, Tim Sweetman wrote: Hmm, sounds like a very different interface.. Currently i call a DBI/DBD::pg function to get a file descriptor, and then i can read(), write() and seek() on it. I used to use Oracle when writing telephony apps for a few years, and that was my first experience of working with blobs. Sounds like i may have become ingrained/infected with that oraclism then! I've always done PostgreSQL stuff in the same way since then. (Been using PostgreSQL for a very long time, but never used to use it with massive fields until v7 era, and it was hard to persuade Vodafone etc to go with an opensource database in the late 90s) So, people are saying it's fine to just create a ByteA or BLOB type and just fill it with 20Mb of binary data. OK then, off to re-write the code :) I can see some disadvantages, in that i can't seek to part-way thru the file and re-write a few bytes, ie random access, but for file attachments in a wiki, that's not going to happen anyway. No.. ta, Toby