On Tue, Oct 10, 2023 at 08:22:55AM +0200, Laurenz Albe wrote: > On Tue, 2023-10-10 at 11:49 +0800, David HJ wrote: > > I am writing to propose an alternative compilation of PostgreSQL that allows > > for a 256-byte identifier length limit, alongside the existing 64-byte version. > > > > Problem: > > The current limit of 63 bytes can be quite restrictive, especially for databases > > that use multi-byte character sets like UTF-8. In such cases, a Chinese character > > takes up 3 bytes, limiting the name to just 21 characters. > > Anyway, you are not the first person to hit the limit, so there is clearly a > pain that some people feel. > > > Proposed Solution: > > I propose that we offer an alternative compilation of PostgreSQL that increases > > the NAMEDATALEN constant to allow for 256-byte identifiers. This would be > > particularly useful for databases that make extensive use of multi-byte character > > sets like UTF-8. > > Wouldn't it be a good solution if we promote the #define to a configure option, > like "./configure --identifier-length-limit=256"? Note that there was some thread recently [1] where the possibility of having some kind of compilation matrix to generate multiple set of binaries with various compile-time values was discussed, so I guess it could fit well with that approach. [1] https://www.postgresql.org/message-id/20230630211153.kbysulcjedxa5ii6@xxxxxxxxxxxxxxxxxx and following messages
