One of the problems faced by the tinderbox of yesteryear is the picking information out of logs, as well as the reliance of one person to interpret the results. With this in mind, I’ve been doing some work to improve accessibility of this data and have produced a tinderbox interface.
A Portage bashrc (based on the original work by Diego Elio Pettenò) collects QA information about builds, and stores it in individual files to make it easier to operate on – eliminating a lot of the need to parse logs.
You’ll notice the interface lists all packages – not just those with a recent build. This allows for a central location to report static analysis information from tools such as repoman and pkgcore-checks. Other lesser-known tools are supported, with experimental reporting of sub-slot candidates and automated dependency checking.
What’s next? I’d like to add ways to find packages beyond the usual category breakdown – such as by maintainer or builds by architecture. There’s more build-time checks to add, and I’m sure there’s other static analysis tools out there too. I don’t personally have the resources to build packages at the scale seen previously, so last but of course not least, more building power is needed. Fortunately, it’s quite easy to collate the tinderbox data from multiple sources so we may be able to ‘crowd-source’ if necessary.
As always, comments/feedback/suggestions welcome.