Re: [MacPorts] #74362: port 2.12.6 does not build on PPC Leopard, Mac OS X 10.5.8, because of error: ‘CURLOPT_NOPROXY’ undeclared (first use in this function)

[MacPorts]

#74362: port 2.12.6 does not build on PPC Leopard, Mac OS X 10.5.8, because of
error: ‘CURLOPT_NOPROXY’ undeclared (first use in this function)
------------------------+-----------------------------
  Reporter:  ballapete  |      Owner:  (none)
      Type:  defect     |     Status:  closed
  Priority:  Normal     |  Milestone:  MacPorts Future
 Component:  base       |    Version:  2.12.6
Resolution:  fixed      |   Keywords:  leopard
      Port:  port       |
------------------------+-----------------------------

Comment (by fhgwright):

 Replying to [comment:6 ballapete]:
 > When will a functioning software be available for Mac OS X 10.5.8,
 Leopard?

 There's already an unreleased fix at the head of the `release-2.12`
 branch, and it's a patchfile in the `MacPorts` port, though it probably
 won't be available to `selfupdate` until it's in a release.

 In the meantime, the easiest way to get the fix is probably:

 {{{
 sudo port build MacPorts
 cd $(port work MacPorts)/MacPorts-2.12.6
 sudo make install
 }}}


 Replying to [comment:4 jmroot]:
 > It's not really unimplemented, it's just passed to libcurl solely in the
 environment, which is unreliable if other threads are modifying it.
 Warning about the combination of no_proxy and fetch_threads > 0 on 10.5
 would be OK.

 It turns out that my fix didn't work right, anyway, since I'd made the
 mistake of assuming that `if (no_proxy)` behaves as expected, when it
 actually seems to be true all the time.  Perhaps the "no-no_proxy" case is
 providing a pointer to a null string, rather than a NULL pointer.  If
 that's the case, I wonder how many other `if (foo)` expressions really
 need to be `if (foo && *foo)`.

-- 
Ticket URL: <https://trac.macports.org/ticket/74362#comment:7>
MacPorts <https://www.macports.org/>
Ports system for macOS