PATCH: fix port::portname dependency

[jberry@…]

Patch enclosed to alter port::portname dependency into port:portname form. This eliminates an issue where TCL sometimes thinks that's a namespace invocation.

The older form port::portname is deprecated, but left in as a short-term transition mechanism. Portfiles will be migrated to the new port:portname form.

[jberry@…]

Patch alters darwinports.tcl and portdepends.tcl

[jberry@…]

Patch alters darwinports.tcl and portdepends.tcl

[jberry@…]

Committed to trunk and release1.

[jmpp@…]

(In reply to comment #0)

The older form port::portname is deprecated, but left in as a short-term transition mechanism. Portfiles will be migrated to the new port:portname form.

James, since such a small numbers of ports were adapted to use this new dependency type during its short infancy in base/, why leave code behind to support an already deprecated syntax? I would vote for removing support for "port::portname" syntax and, if possible, have portindex(1) fail in parsing any Portfile that uses it, so that it cathes our eyes for fixing.

Thoughts?

-jmpp

[jberry@…]

Hey Juan,

The "support code" left behind is just two characters in a regexp, so it's not killing us from that standpoint. I left the support in because I didn't want to force everybody to update in sync.

But I'm all for taking the legacy support out soon...within weeks, to prevent anybody from relying on it accidentally. I don't think it needs to come out before release1; since it's deprecated, it's not something anybody should rely on, and it's not something that's supported as part of the spec for a release1 compatible portfile.

Make sense?

-jdb

[jmpp@…]

(In reply to comment #5)

Hey Juan,

The "support code" left behind is just two characters in a regexp, so it's not killing us from that standpoint. I left the support in because I didn't want to force everybody to update in sync.

But I'm all for taking the legacy support out soon...within weeks, to prevent anybody from relying on

it

accidentally. I don't think it needs to come out before release1; since it's deprecated, it's not

something

anybody should rely on, and it's not something that's supported as part of the spec for a release1 compatible portfile.

Make sense?

-jdb

Hey James! Yes, your explanation makes sense. But my point is that I see no need in leaving the support in if no one, except Ole and you, ever used the port::foo syntax. And both you guys already uprgaded to the new one, so... But since this "support" is only a thing of two characters in a regexpr then I guess it's not too bad, but I wouldn't want to see anyone using the "old" syntax and not realizing the mistake (mainly because the Portfile will parse correctly), only to be bitten later on by user reports of failure. Just my thougths. Opinions anyone else?

-jmpp