On Apr 13, 2007, at 22:34, source_changes@macosforge.org wrote:
Modified: trunk/base/doc/portfile.7 (23967 => 23968) --- trunk/base/doc/portfile.7 2007-04-14 01:54:40 UTC (rev 23967) +++ trunk/base/doc/portfile.7 2007-04-14 03:34:33 UTC (rev 23968) @@ -1331,6 +1331,26 @@ .br .Sy Example: .Dl default_variants +ssl +tcpd +.It Ic unversal_variant +When using MacPorts on MacOS X, a universal variant is defined and +the default behavior is to configure ports with universal flags
Wait... I thought the default behavior was for the +universal variant to be available to all ports, but that you had to select it if you wanted to use it.
On Apr 15, 2007, at 11:53 AM, Ryan Schmidt wrote:
On Apr 13, 2007, at 22:34, source_changes@macosforge.org wrote:
Modified: trunk/base/doc/portfile.7 (23967 => 23968) --- trunk/base/doc/portfile.7 2007-04-14 01:54:40 UTC (rev 23967) +++ trunk/base/doc/portfile.7 2007-04-14 03:34:33 UTC (rev 23968) @@ -1331,6 +1331,26 @@ .br .Sy Example: .Dl default_variants +ssl +tcpd +.It Ic unversal_variant +When using MacPorts on MacOS X, a universal variant is defined and +the default behavior is to configure ports with universal flags
Wait... I thought the default behavior was for the +universal variant to be available to all ports, but that you had to select it if you wanted to use it.
Hello Ryan, This is what I meant, or at least tried to express. +universal will be available to all ports unless they specify "universal_variant no". And universal variant code will have no effect (except listing this variant as available) unless users select +universal. The default behavior *of the universal variant* is to configure with universal flags. Please feel free to correct the wording to make the man page easier to understand. Paul
On Apr 14, 2007, at 22:04, Paul Guyot wrote:
On Apr 15, 2007, at 11:53 AM, Ryan Schmidt wrote:
On Apr 13, 2007, at 22:34, source_changes@macosforge.org wrote:
Modified: trunk/base/doc/portfile.7 (23967 => 23968) --- trunk/base/doc/portfile.7 2007-04-14 01:54:40 UTC (rev 23967) +++ trunk/base/doc/portfile.7 2007-04-14 03:34:33 UTC (rev 23968) @@ -1331,6 +1331,26 @@ .br .Sy Example: .Dl default_variants +ssl +tcpd +.It Ic unversal_variant +When using MacPorts on MacOS X, a universal variant is defined and +the default behavior is to configure ports with universal flags
Wait... I thought the default behavior was for the +universal variant to be available to all ports, but that you had to select it if you wanted to use it.
This is what I meant, or at least tried to express. +universal will be available to all ports unless they specify "universal_variant no". And universal variant code will have no effect (except listing this variant as available) unless users select +universal. The default behavior *of the universal variant* is to configure with universal flags.
Please feel free to correct the wording to make the man page easier to understand.
I see what you meant now. :) So, these changes you're making... are they the solution to the problem I reported when I want to define my own universal variant? Should I be (as soon as released MacPorts supports it) using "universal_variant no" before defining my own universal variant?
On Apr 15, 2007, at 12:32 PM, Ryan Schmidt wrote:
I see what you meant now. :)
So, these changes you're making... are they the solution to the problem I reported when I want to define my own universal variant? Should I be (as soon as released MacPorts supports it) using "universal_variant no" before defining my own universal variant?
No. You just define your own universal variant with: variant universal { # Your code goes here. } As a matter of fact, if you do: variant universal { # some code } universal_variant no You'll end up having no universal variant in your portfile. But this is an undocumented behavior. Paul
On Apr 14, 2007, at 22:41, Paul Guyot wrote:
So, these changes you're making... are they the solution to the problem I reported when I want to define my own universal variant? Should I be (as soon as released MacPorts supports it) using "universal_variant no" before defining my own universal variant?
No. You just define your own universal variant with:
variant universal { # Your code goes here. }
Ok, that's what I'm doing (in sleepwatcher), but it's resulting in weirdness: "port info" shows two "universal" variants; see: $ port info sleepwatcher | head -n 1 sleepwatcher 2.0.4, Revision 2, sysutils/sleepwatcher (Variants: universal, server, universal) (This is with my ports tree indexed by portindex from macports trunk as of a few moments ago.)
On Apr 15, 2007, at 1:42 PM, Ryan Schmidt wrote:
Ok, that's what I'm doing (in sleepwatcher), but it's resulting in weirdness: "port info" shows two "universal" variants; see:
$ port info sleepwatcher | head -n 1 sleepwatcher 2.0.4, Revision 2, sysutils/sleepwatcher (Variants: universal, server, universal)
(This is with my ports tree indexed by portindex from macports trunk as of a few moments ago.)
Good catch! The actual behavior is unchanged (redefining a variant basically overrides the previous implementation), but indeed the variant was listed twice in portinfo. This is fixed in r24100. However, please note that you'll see two variants for sleepwatcher until PortIndex is generated with r24100 update (until next release, I believe). Paul
On Apr 16, 2007, at 06:21, Paul Guyot wrote:
Ok, that's what I'm doing (in sleepwatcher), but it's resulting in weirdness: "port info" shows two "universal" variants; see:
$ port info sleepwatcher | head -n 1 sleepwatcher 2.0.4, Revision 2, sysutils/sleepwatcher (Variants: universal, server, universal)
(This is with my ports tree indexed by portindex from macports trunk as of a few moments ago.)
Good catch! The actual behavior is unchanged (redefining a variant basically overrides the previous implementation), but indeed the variant was listed twice in portinfo. This is fixed in r24100. However, please note that you'll see two variants for sleepwatcher until PortIndex is generated with r24100 update (until next release, I believe).
Looks good now in trunk. Thanks!
participants (2)
-
Paul Guyot
-
Ryan Schmidt