Обсуждение: proposal: new contrib module - session variables
Hello all, session variables are missing feature long time. Currently nobody works on package or SQL/PSM module implementation so there is some gap.Currently best know solution are session variables based on plper hash tables or Joe Conway's C implementation. I propose include Joe's code into contrib set http://www.joeconway.com/sessfunc.tar.gz It's important mainly for window's users, that has problems with own compilation. Objections? Regards Pavel Stehule
Pavel Stehule wrote: > Hello all, > > session variables are missing feature long time. Currently nobody > works on package or SQL/PSM module implementation so there is some > gap.Currently best know solution are session variables based on plper > hash tables or Joe Conway's C implementation. > > I propose include Joe's code into contrib set > http://www.joeconway.com/sessfunc.tar.gz It's important mainly for > window's users, that has problems with own compilation. > > Objections? > > The URL fails, so I can't comment on Joe's work. But I seriously doubt that any contrib module is really the right way to go about doing session variables. If we're going to have them then they need to be properly builtin, and available to all PLs. cheers andrew
2008/6/9 Andrew Dunstan <andrew@dunslane.net>: > > > Pavel Stehule wrote: >> >> Hello all, >> >> session variables are missing feature long time. Currently nobody >> works on package or SQL/PSM module implementation so there is some >> gap.Currently best know solution are session variables based on plper >> hash tables or Joe Conway's C implementation. >> >> I propose include Joe's code into contrib set >> http://www.joeconway.com/sessfunc.tar.gz It's important mainly for >> window's users, that has problems with own compilation. >> >> Objections? >> >> > > The URL fails, so I can't comment on Joe's work. But I seriously doubt that > any contrib module is really the right way to go about doing session > variables. If we're going to have them then they need to be properly > builtin, and available to all PLs. > probably Joe's server is out currently :(. Joe's implementation was simply - hash table stored in session context. Pavel > cheers > > andrew >
Pavel Stehule wrote: >> The URL fails, so I can't comment on Joe's work. But I seriously doubt that >> any contrib module is really the right way to go about doing session >> variables. If we're going to have them then they need to be properly >> builtin, and available to all PLs. > > probably Joe's server is out currently :(. > > Joe's implementation was simply - hash table stored in session context. Sorry -- I forgot to restart apache after upgrading my kernel yesterday :-( I don't have a strong opinion either way regarding including this in contrib or elsewhere, but if there is sufficient interest I'll find the time to update it for cvs head as well as ensure it represents the latest and greatest that I have available. I've used and improved it in minor ways since that tarball was put up. Joe