port callback example

[James Gregurich]

I'm not opposed to donating some time to look at this issue down the road, but I need to get this task over with so I can get back to my other project for which I am under the gun to get finished in the near future. 

BTW: Just so you know why I'm doing this work:

My company, Markzware, is a mac development shop. We sell a number of desktop products, and we use macports to manage our opensource library dependencies. We have some ideas for iphone/ipad apps but we need to extend our infrastructure to handle the additional requirements for the iOS platform. My options are 


1) ditch macports for something else
2) donate the time to extend macports to handle iOS
3) keep macports and then manually manage iOS dependencies. 

I've chosen 2) as this is the best option for the long-run. Given the popularity of iOS development these days...I figure we're not the only ones who would find this functionality useful.


-James

On Feb 8, 2011, at 3:04 PM, Ryan Schmidt wrote:

> 
> On Feb 8, 2011, at 16:55, James Gregurich wrote:
> 
>> Correct me if my understanding is wrong, but my analysis is that if I overrode configure in port using the muniversal portgroup, I would have to supply a new version of the entire configure function.  Is this incorrect?
>> 
>> I also note that pre-configure and post-configure are only called once...not once-per-arch.
>> 
>> As I see it, my only opportunity to patch the configure action is my supplying configure.cmd and configure.args. I don't see a way to add in code to configure.cmd that would output data to user-config.jam.
>> 
>> 
>> Is my analysis wrong? If not, what is the recommended course of action if adding additional call-backs is undesirable?
> 
> One thing that has bothered me about muniversal is that the arch-specific directory is created at the beginning of its configure phase. This means that if I write a pre-configure block, that's too early to do anything in those arch-specific directories, because they don't exist yet. It would make more sense to me if muniversal would make those copies of the original directory in post-patch, say. That way, by pre-configure time, they'd all be there and you could write a pre-configure block in your port to do something with them. But I've never tried changing this behavior of muniversal to see if it breaks something.
> 
>