Обсуждение: What's that query limit again?
Greetings, The query limit is 8K right? What about the size of rows returned by a select? If it is 8K too, then how might I be sure I can get all the results of a row that has several "text" fields? The user may type more than 8K total across all the fields... I'm trying to avoid large objects. Thanks, Matthew
On Thu, 25 Feb 1999, Matthew Hagerty wrote: > Greetings, > > The query limit is 8K right? What about the size of rows returned by a > select? If it is 8K too, then how might I be sure I can get all the > results of a row that has several "text" fields? The user may type more > than 8K total across all the fields... > > I'm trying to avoid large objects. There's two 8k limits, one the size of the query, the other the block size in the table. The latter is changeable (at compile time) upto 32K with varying levels of success. The size of a returned result is only limited to the resources of the server, and in some cases the client. Peter -- Peter T Mount peter@retep.org.uk Main Homepage: http://www.retep.org.uk PostgreSQL JDBC Faq: http://www.retep.org.uk/postgres Java PDF Generator: http://www.retep.org.uk/pdf
Matthew Hagerty <matthew@venux.net> wrote: > The query limit is 8K right? This 8K limit is going to kill Postgres, IMHO. I really like Postgres but I am going to have to move to some other database platform so that I can populate a web site from a database without having to jump through hoops to split text into 4K blocks. > I'm trying to avoid large objects. I don't see the need for them. My opinion is that the backend should hide large objects from the user. It's all spherical objects to me! -- Mark Himsley. Technical Director, Clockwork Web.
Matthew Hagerty <matthew@venux.net> writes: > The query limit is 8K right? What about the size of rows returned by a > select? There's an 8K limit on the text of a submitted query (I hope to eliminate that soon, but probably not in time for 6.5). As far as I can think offhand, there is no a-priori limit on the size of a query result in either columns or rows. (Well, I notice a 16-bit column count in T messages, so the FE/BE protocol will break at 32K or 64K columns in a result, but hopefully no one will ever run into that...) Have you tried it? An experiment would've taken less time than querying the list, I'm sure. regards, tom lane