[73039] trunk/dports/_resources/port1.0/group/qt4-1.0.tcl

[Joshua Root]

On 2010-11-6 05:16 , Michael Dickens wrote:
> On Nov 5, 2010, at 11:26 AM, Joshua Root wrote:
>> Yes, phonon is evaluating qt_arch_types prematurely. So don't do that.
> 
These Portfiles (e.g., phonon) were working just fine using an SVN
verion of 'port' from maybe 3 weeks ago.  So, was it wrong of me to use
this 'options' variable in the Portfile context instead of a stage
context?  If so, is the 'options' variables concept documented somewhere
so that I can understand better how/when to use them?  Any idea what
changed in 'port' recently to make this 'options' variable evaluate
prematurely?  I have no idea where to even start looking to try to debug
when/where this happens. - MLD

Sorry if that came off as brusque, it was past my bedtime. :-)

I don't see how the ports could have been working earlier unless they
were doing something different. Before the change I made to the
portgroup, the qt_arch_types variable didn't even exist until the
configure phase.

The option is being evaluated at time of use, as expected. The reason
you weren't getting the universal archs is that the proc used to
evaluate it checks for the existence of a universal variant, which was
not yet declared when you read the variable.

If you were to add your own proc to the portgroup to set up
qt_arch_types, it would have to make the same check for a universal
variant and would thus have the same limitations.

- Josh