Re: [PATCH v1 15/20] node_device_udev: Pass the driver state as parameter in preparation for the next commit

On Tue, Apr 23, 2024 at 09:10 AM +0100, Daniel P. Berrangé <berrange(a)redhat.com&gt; wrote: > On Tue, Apr 23, 2024 at 10:03:35AM +0200, Marc Hartmayer wrote: >> On Fri, Apr 19, 2024 at 02:23 PM -0500, Jonathon Jongsma <jjongsma(a)redhat.com&gt; wrote: >> > On 4/19/24 9:49 AM, Marc Hartmayer wrote: >> >> It's better practice for all functions called by the threads to pass the driver >> >> via parameter and not global variables. Easier to test and cleaner. >> >> >> >> […snip…] >> >> >> >> >> >> >> static int >> >> -udevProcessPCI(struct udev_device *device, >> >> +udevProcessPCI(virNodeDeviceDriverState *driver_state, struct udev_device *device, >> >> virNodeDeviceDef *def) >> > >> > While there are exceptions, the general coding style is to have a single >> > argument per line for function definitions. >> >> Okay. BTW, why is there no .clangformat configuration available for >> Libvirt? :/ > > There is no combination of clangformat settings that can match > libvirt code style. If we were startnig again from scratch we > would of course want to match a defined clangformat style, > but unless we're willing to bulk reformat the codebase we are > stuck with what we have. Hmm, is the downside of doing a full codebase reformat greater than always doing the code formatting manually? Probably it is, otherwise it would have already be done :) If we would have such a big commit we could list it in a `.git-blame-ignore-revs` file so it will ignored by git blames.