On Wed, 09/21 09:29, Eric Blake wrote: > On 09/21/2016 09:01 AM, Fam Zheng wrote: > > On Wed, 09/21 15:44, Greg Kurz wrote: > >> On Wed, 21 Sep 2016 06:35:04 -0700 (PDT) > >> no-re...@patchew.org wrote: > >> > >>> Hi, > >>> > >>> Your series failed automatic build test. Please find the testing commands > >>> and > >>> their output below. If you have docker installed, you can probably > >>> reproduce it > >>> locally. > >>> > >> > >> Heh... of course patchew doesn't know about Stefan's series. :) > > > > Usually those patchsets fail to apply and patchew keeps quiet. This series > > is > > the first victim of a exception, seemingly. :) > > While we're at it, is there a way to make the error messages include the > last 15 lines or so of the build log first, and then the full build log? > Usually, the failure is easy to identify from the tail of the message, > but having to scroll to the bottom past all the successful portion of > the log is slower than seeing the tail up front. At the same time, the > full log is useful for the cases where the failure occurs earlier than > the last 15 lines.
Scrolling to bottom is quite easy in any email client, while more sections in a long email demand extra effort to manage and read. I doubt adding a "tail of log" section just before the full log helps us much. I'd rather keep it simple and stupid. > > > > >> > >> Is there an appropriate way to avoid complaints when sending a patchset > >> that > >> isn't based on QEMU master ? > > > > Currently we don't, but we can invent one. Like: > > > > Base: http://github.com/someone/qemu branch > > I like that idea (these patches apply from that point; then it is up to > patchew whether to bother chasing down that point or just skipping the > series) > > > > > for the "base on maintainer tree" case, or just inform patchew to "pull" > > instead of apply: > > > > Git: http://github.com/someone/qemu topic > > Maybe Pull: instead of Git:, but again a nice idea. I'm adding them to my list. > > > > > > for the "base on the other series" case. > > > > Supporting this is not complicated, the tricky thing is to remember to put > > that > > magic in cover letters every time a series doesn't base on master. > > > > Any thoughts on the harder question? :) > > Which one was the harder question? How to get people to remember to use > magic that patchew can recognize? Yes. :) > > -- > Eric Blake eblake redhat com +1-919-301-3266 > Libvirt virtualization library http://libvirt.org >