Bug 71177
| Summary: | repo-copy request: [MAINTAINER] databases/postgresql-server: update to 7.4.5 and client/server split | ||||||
|---|---|---|---|---|---|---|---|
| Product: | Ports & Packages | Reporter: | Palle Girgensohn <girgen> | ||||
| Component: | Individual Port(s) | Assignee: | Ade Lovett <ade> | ||||
| Status: | Closed FIXED | ||||||
| Severity: | Affects Only Me | CC: | portmgr | ||||
| Priority: | Normal | ||||||
| Version: | Latest | ||||||
| Hardware: | Any | ||||||
| OS: | Any | ||||||
| Attachments: |
|
||||||
Responsible Changed From-To: freebsd-ports-bugs->ade As sent in email to submitter and portmgr@, since I opened the Pandora's Box here, it's only right that I take point on getting this into the tree, though it will not happen until 5.3-REL is out the door. Palle Girgensohn wrote:
> patch Mk/bsd.port.mk with stuff shamelessly stolen from the MySQL port
Besides needing entries for MOVED and UPDATING, *please do not check
WITH_x variables in bsd.port.mk*.
-Oliver
--On tisdag, augusti 31, 2004 10.28.44 +0200 Oliver Eikemeier <eikemeier@fillmore-labs.com> wrote: > Palle Girgensohn wrote: > >> patch Mk/bsd.port.mk with stuff shamelessly stolen from the MySQL port > > Besides needing entries for MOVED and UPDATING, *please do not check > WITH_x variables in bsd.port.mk*. Hmm, OK. I merely copied the myslq behaviour, it been around a while so I thought it would be OK. So, you mean, if the users wants a non-default version of postgresql, he has to install that version of postgresqlNN-client, and it will discovered automatically. Sure, why not. I'll send a patch. /Palle Palle Girgensohn wrote: > --On tisdag, augusti 31, 2004 10.28.44 +0200 Oliver Eikemeier > <eikemeier@fillmore-labs.com> wrote: > >> Palle Girgensohn wrote: >> >>> patch Mk/bsd.port.mk with stuff shamelessly stolen from the MySQL port >> >> Besides needing entries for MOVED and UPDATING, *please do not check >> WITH_x variables in bsd.port.mk*. > > Hmm, OK. I merely copied the myslq behaviour, it been around a while so > I thought it would be OK. So, you mean, if the users wants a > non-default version of postgresql, he has to install that version of > postgresqlNN-client, and it will discovered automatically. Sure, why > not. I'll send a patch. No. What I would prefer (personal preference) is that a non-default version of postgresql has to be specified in /etc/make.conf, and no autodetection takes places. The point here is that every != spawns a process, even when no build is intended (like INDEX builds, or other tools operating on the ports tree). Anyway, what I wanted to say is that WITH_x is reserved for user -> port communication and must not be used in bsd.port.mk, I've been already bitten by WITH_MYSQL_VER once. I don't believe the MySQL stuff in bsd.port.mk makes a pretty good example that should be followed... -Oliver --On tisdag, augusti 31, 2004 12.36.46 +0200 Oliver Eikemeier <eikemeier@fillmore-labs.com> wrote: > Palle Girgensohn wrote: > >> --On tisdag, augusti 31, 2004 10.28.44 +0200 Oliver Eikemeier >> <eikemeier@fillmore-labs.com> wrote: >> >>> Palle Girgensohn wrote: >>> >>>> patch Mk/bsd.port.mk with stuff shamelessly stolen from the MySQL port >>> >>> Besides needing entries for MOVED and UPDATING, *please do not check >>> WITH_x variables in bsd.port.mk*. >> >> Hmm, OK. I merely copied the myslq behaviour, it been around a while so >> I thought it would be OK. So, you mean, if the users wants a >> non-default version of postgresql, he has to install that version of >> postgresqlNN-client, and it will discovered automatically. Sure, why >> not. I'll send a patch. > > No. What I would prefer (personal preference) is that a non-default > version of postgresql has to be specified in /etc/make.conf, and no > autodetection takes places. > > The point here is that every != spawns a process, even when no build is > intended (like INDEX builds, or other tools operating on the ports tree). > > Anyway, what I wanted to say is that WITH_x is reserved for user -> port > communication and must not be used in bsd.port.mk, I've been already > bitten by WITH_MYSQL_VER once. I don't believe the MySQL stuff in > bsd.port.mk makes a pretty good example that should be followed... Thinking about it, I think you're right about this. I'll consider makeing some modifications... /Palle Hi! I've been sitting on this for some time, polishing it, over the port freeze and apparently beyond. This mega patch upgrades to 7.4.6 | 7.3.8 and refines the splitting into -client & -server ports. The previous patches attached to this PR can be skipped in favour for these new ones. It also modifies all ports depending on postgresql. With so many ports changed, there are bound to be bugs, but most mods are pretty simple, so hopfully they won't bite too hard. Instead of including it all here, I've put it on a web server: http://pingpong.net/postgresql-split.tar.bz2 There's a README inside that hopefully explains what to do. Regards, Palle State Changed From-To: open->closed Moving to ports/73843 for repocopy, patch, and exp build run. Hi! Perhaps we should request a repo copy of postgresql74-server -> postgresql80-server at the same time? I have a patch ready this weekend. /Palle Hi Ade! I rerolled the diffs against the repo-copied postgresql-ports, and remade the entire tarball. It is in http://pingpong.net/postgresql-split-1.1.tar.bz2. I have tried to handle the PORTREVISIONs properly. By properly, I mean that they are bumped if postgresql is not optional. If the user has to add a switch "WITH_PGSQL=yes" or similar (often the case, with MySQL as default), I let the PORTREVISION be. Dunno if it is the correct procedure, but it is what I would expect, I think, at least if I didn't use postgresql... Do I need bump PORTREVISION for all ports? They will not really build different, just the package dependencies will change. --On fredag, november 19, 2004 15.15.23 +0100 Palle Girgensohn <girgen@pingpong.net> wrote: > Hi Ade! > > Just checked the patchset for the postgresqlNN-server ports, perhaps it > was not quite as high quality as I thought. I'll fix a new set of > everything asap tonight, so you don't have to search patches in a bunch > of emails. > > In the mean time, can you please set BROKEN or IGNORE on the repo copied > ports? > > Regards, > Palle > > --On Tuesday, November 16, 2004 21:19:44 -0800 Ade Lovett > <ade@FreeBSD.org> wrote: > >> Hi Palle, >> >> Just FYI, my requested repo-copies from >> postgresql73->postgresql73-server, and postgresql7->postgresql74-server >> have been complete, and are now in the tree. >> >> If you can send me over a now somewhat easier patchset for the remainder, >> I can push to get this tested and committed. >> >> Cheers, >> -aDe > > > |
- Update to 7.4.5 - Split port into a -server and a -client part - Handle slave ports better, esp. with older server versions This upgrade needs a repo copy from postgresql7 and modifies bsd.port.mk I'll be sending more stuff for postgresql 7.2, 7.3 and 8.0 tomorrow, and also patches for all ports requiring postgresql-client. Perhaps best to commit it all at once? Instructions for how to upgrade (also in the README inside the tarball): patch Mk/bsd.port.mk with stuff shamelessly stolen from the MySQL port bsd.port.mk.diff Then, repo-copy: postgresql7 -> postgresql74-server (+) patch the new port accordingly with patch: postgresql74-server.diff cvs remove files (or skip repo copy) postgresql74-server/pkg-install postgresql74-server/pkg-plist postgresql74-server/files/pkg-message.server postgresql74-server/files/post-install-notes cvs add files: postgresql74-server/pkg-message-client postgresql74-server/pkg-message-contrib postgresql74-server/pkg-message-jdbc postgresql74-server/pkg-message-plperl postgresql74-server/pkg-message-pltcl postgresql74-server/pkg-message-server postgresql74-server/pkg-message-tcltk postgresql74-server/pkg-plist-client postgresql74-server/pkg-plist-jdbc postgresql74-server/pkg-plist-plperl postgresql74-server/pkg-plist-pltcl postgresql74-server/pkg-plist-server postgresql74-server/pkg-plist-tcltk postgresql74-server/pkg-install-server (+) new port: postgresql74-client.shar modified slave ports: p5-postgresql-plperl.diff postgresql-contrib.diff postgresql-docs.diff postgresql-jdbc postgresql-pltcl.diff postgresql-tcltk.diff cvs remove \ p5-postgresql-plperl/pkg-plist \ p5-postgresql-plperl/pkg-message \ postgresql-contrib/pkg-plist \ postgresql-contrib/pkg-message \ postgresql-docs/pkg-plist \ postgresql-jdbc/pkg-message \ postgresql-jdbc/pkg-plist \ postgresql-pltcl/pkg-plist \ postgresql-pltcl/pkg-message \ postgresql-tcltk/pkg-plist (+) the files pkg-plist, files/post-install-notes and files/pkg-message.server don't need to be repo-copied, perhaps easiest to copy them with the rest and then remove? the file pkg-install should really be repo-copied -> pkg-install-server, but perhaps not worth the effort?