Re: Universal build of cranky open source lib with a configure script
Re: Universal build of cranky open source lib with a configure script
- Subject: Re: Universal build of cranky open source lib with a configure script
- From: Bill Bumgarner <email@hidden>
- Date: Fri, 26 Jan 2007 19:04:25 -0800
On Jan 26, 2007, at 5:12 PM, Bill Northcott wrote:
No it does not. As long as the one of the arch options is runnable
on the compiling host the autoconf tests should work just nice and
do wat they are supposed to do. Some one else pointed out the few
pitfalls - endianness etc. Most packages are not concerned about
CPU architecture only the availability of appropriate libraries and
tools and that is what the autoconf tests for.
Ticking time bomb.
I have seen -- and had to fix -- far too many issues related to
assuming that code is working in this kind of situation to be
comfortable with such a solution. It just isn't worth it; creating
an Xcode project that builds a proper universal binary is a known
quantity, not that hard, and a heck of a lot more productive/fun than
trying to debug endian, linking, ABI or other weirdness in the final
bomb run of a project.
The worst kinds have been the ones that yielded non-portable data
archives that should have been portable. Nothing like having to
figure out the architecture of the original machine used to build the
software to recover a bunch of data that was archived/unarchived
incorrectly.
While I completely agree that such a casual approach to building
random open source is likely fine for hobbyist use, I would not rely
upon it for production use nor could I recommend that anyway else do
so. And I have seen enough pain caused by such assumptions regarding
autoconf that I feel this is a horse worth a bit of flogging.
b.bum
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Xcode-users mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden