I am becoming convinced that blogging is evil (“he writes in his blog one afteroon.”). I used to have a normal life. Then I decided I might be just un-lazy enough to keep a blog on my site. Then Planet Gentoo emerged (hah!) and my tiny Gentoo section was being fed over there. All these other devs, some I’ve known for years, start posting, showing what they’re working on, where they’re going, what they’re doing to make Gentoo great, so I feel the pressure to start posting. But what to post about? Guess I’ll have to crack and do some real work – and pow, I’m a madman on bugzilla, being all overproductive and what not so I have something to talk about in here. I start making up work to do just to have a conversation with myself and you, my unknown audiance. And hence, blogging is evil, because it promotes productivity. It’s in the Book as one of the signs of the apocalypse, I swear.
module-build gotchas – OK, so I spoke too soon yesterday. A few bugs have cropped up now that the eclass is attempting to intelligent checking for Module::Build scripts, but it’s not my fault, I swear. The problem is that we have no means that I know of in portage to bootstrap a package only on the first install – that is, if the package has never been installed, install it this way, otherwise install it this way. Wouldn’t be an issue except that Module::Build has an optional dependancy on ExtUtils::ParseXS – no problem, except the latest version of E::P contains a Build.PL, which invokes our eclass friend to complain and die because M::B isn’t installed, which complains because E::P wasn’t installed. Sadly, the easiest solution for the time being is to remove the optional dep, but it’s little junk like that brings a good eclass mod down.
No other news – namely on account of the module-build work through the last day, no time to work on other bugs, like say the huge tracker bug for perl-core modules vs ebuild modules version compliance. Though I have been playing with g-cpan, trying to put it through its paces, only found one glitch so far out of a lot of emerging (CPANPLUS gets some conf values wrong that you have to reset on first usage, but still, pretty cool. I’d still like to see a day where g-cpan is the user interface, and we have some nifty plugin to portage that can do the same thing on the fly for commited ebuilds (ie, G-DEPEND=”module” and portage will automagically emerge it instead of us having a ton of ebuilds in the tree that are all skeletons).
I’m sure there was something else, probably nifty keen, to say, though I don’t know what it could have been at this point…