[cairo] patches for configure errors under MinGW, OS X deprecation warnings

[Ryan Schmidt]

On Feb 8, 2015, at 7:31 PM, Bryce Harrington wrote:

> On Sun, Feb 08, 2015 at 01:10:16PM +0100, Andrea Canciani wrote:
>> On Sun, Feb 8, 2015 at 11:17 AM, Ryan Schmidt wrote:
>>> Would it be possible to keep the previous code for OS X 10.4 and use the
>>> new code on OS X 10.5 and newer?
>> 
>> It is possible to do that, but testing on 10.4 is becoming pretty hard (I
>> have no such environment and I'm afraid that Mozilla is not testing anymore
>> on such an old OS). In the bugreport the agreement seemed to be that 10.4
>> is too old to be supported. Its removal would also allow for a significant
>> cleanup, as 10.4 is missing blend operators.
>> 
>> I am unsure what the best course is... Chris, Bryce, what are your opinions
>> on this?
> 
> On the Linux distro side, LTS releases typically are supported no more
> than 5 years, so Ubuntu's 8.04 Hardy Heron released in 2008 was no
> longer supported as of 2013.  If someone asked about a Cairo problem
> that was specific to Hardy and not an issue as of Lucid (10.04), I'm
> fairly sure we'd close the bug with a recommendation to upgrade.
> 
> Wikipedia shows OS X 10.4 released in 2003 with final point release in
> 2007.  OS X 10.5 was released in 2007, with its last release in 2009.
> I suppose OS X is a bit different than a Linux distro in that users
> actually have to pay money for the upgrades, but that's sort of the
> deal with the devil they made picking proprietary software.
> 
> In Inkscape we require 10.5 or newer, and I think we've required that
> for a while.  On the Linux side we support LTS's from 2012 and newer;
> maybe a bit older for Fedora, I forget.
> 
> I don't know if we have an established policy in Cairo, but setting a
> expectation of an OS from 2008 or newer seems adequately conservative
> for an established system library.  Seven years is longer than most
> LTS's and probably beyond the typical lifespan of most consumer
> hardware.
> 
> So, my opinion is yes, I think we can bid 10.4 adieu.

Understand that it's not just a matter of a user choosing not to upgrade the OS. There are many Macs which cannot be upgraded past 10.4 (as there are for every other version of Mac OS and OS X, of course). These are certainly old machines by now, but we do still have a handful of MacPorts users using 10.4. I would not be surprised if those users also have newer Macs running newer versions of OS X, but it is valuable to be able to still use older machines for something, and MacPorts helps users do that.

I am the maintainer of cairo in MacPorts, where, with the recent cairo 1.14.0 update, I finally made the long-desired change to default cairo to enabling quartz, in addition to X11. Now, if quartz font support won't work on 10.4 anymore, I have to modify the port again, to either disable all of quartz on 10.4, or to just disable quartz font support on 10.4. If quartz font support is not used by anyone, that's fine, but if it is, then there will be build failures in those apps. I could also just maintain forever (or until MacPorts itself no longer works on 10.4) a patch to cairo to make quartz font compatible with 10.4 again by reverting this commit, but I would prefer not to have to maintain patches.