[28240] trunk/dports/PortIndex

[cssdev at mac.com]

On Aug 25, 2007, at 5:16 PM, Ryan Schmidt wrote:

> On Aug 25, 2007, at 15:56, Juan Manuel Palacios wrote:
>
>> On Aug 25, 2007, at 4:34 PM, Ryan Schmidt wrote:
>>
>>> On Aug 25, 2007, at 14:36, source_changes at macosforge.org wrote:
>>>
>>>> Total number of ports parsed:	4194
>>>> Ports successfully parsed:	4193	
>>>> Ports failed:			1
>>>>
>>>>
>>>> Failed to parse file palm/hrsh2csv/Portfile: Error evaluating  
>>>> variants
>>>
>>> What's going on here? I don't get this error when creating the  
>>> portindex locally.
>>
>> or directory
>>
>> The variant's main body is being evaluated and naturally the  
>> tarball has not been extracted yet, so there's no Makefile to  
>> edit. I'm guessing wrapping up that reinplace in the platform  
>> block with a {pre,post}-{configure,build} hook would clear the  
>> error. Chris, could you please look into it?
>
> Bingo. You've nailed it. Chris, I've reopened bug #12555 and  
> attached a new patch to it which I think will fix the problem.

Your fix seems to work fine. I lost access to a PowerPC when my  
PowerBook died. :)

I tried to create a platform variant for the change, but I did so a  
bit hastily. Something like "post-extract-append" won't work for the  
variant, but I could get a similar result using "pre-configure"  
within the variant. Ideally, the code should detect endian ordering  
in a more dynamic manner, but that would take more effort to peel  
apart the code. The package should reflect the endian-specific nature  
of the code, IMO. Could someone please verify the latest patch in the  
bug on a PowerPC Mac?

Thanks,
Chris