Problem with gcc4.7 and call_once

[Jeremy Huddleston Sequoia]

On Aug 9, 2013, at 13:17, Daniel J. Luke <dluke at geeklair.net> wrote:

> On Aug 9, 2013, at 4:12 PM, Ryan Schmidt <ryandesign at macports.org> wrote:
>> On Aug 9, 2013, at 15:06, "Brian D. McGrew" <brian at visionpro.com> wrote:
>>> Well I had to port install gcc48 to build gcc-4.8.1 from source.  Using
>>> /usrŠgcc didn't get me there.  So installing port gcc48, building
>>> gcc-4.8.1 from source in ~ and installing over the top of port installed
>>> gcc worked (backing up /opt/local of course).  Circular for sure, but
>>> effective.
>> 
>> And completely impossible to implement in that fashion in MacPorts. A port cannot depend on itself.
> 
> not impossible, just impractical (I think we've had other ports that need themselves to bootstrap).

Yes, see apple-gcc42 on Tiger as an example (the +bootstrap variant).  apple-gcc42 is needed to build dependencies of apple-gcc42 on Tiger.

We really shouldn't be doing that unless there is no other way.


-------------- next part --------------
A non-text attachment was scrubbed...
Name: smime.p7s
Type: application/pkcs7-signature
Size: 4145 bytes
Desc: not available
URL: <http://lists.macosforge.org/pipermail/macports-users/attachments/20130809/29115665/attachment.p7s>