I just released a new version of Linux::USBKeyboard, now with Devel::CheckLib (but I have to say I'm a little sad about adding 10k to my dist (4.6k of which is pod) to check whether a linux user has a compiler and libusb (though I'll admit that the inc::InlineBuild code is suboptimal too (because who has time to write better build tools?).))
So, am I the only module author who thinks that the automated testing procedures have gotten somewhat removed from actual users' computers? Or possibly actual users' computers have gotten really stale.
In this particular case, the installer should really just figure out what represents the libusb-dev package on the current distro and try to install that. But wait, the user shouldn't be root, so we have to hope they can sudo aptitude, etc, etc. While we're at it, we might as well roll a package and install that because perhaps they'll want to uninstall it without breaking any dependencies, etc, etc.
But who has time to write build tools?
What do Perl users want from their CPAN modules? I frequently hear complaints about CPAN being hard to use, yet we call CPAN our killer app. Perhaps the new crop of building-from-source RedHat users will have more to say about that.
But, really. Who has time to write build tools? It seems to me that we've hit an uncanny valley where the tools are getting stale and vendor support is getting behind, but not so much that existing users can't muddle through. I suspect this might mean that new users are giving up, particularly when it comes to using new code. But, the "good enough" means that nothing hurts enough for anyone to invest in making it better.
