On Thu, May 10, 2018 at 10:53:47PM +0200, Matthijs Kooijman wrote: > Hi Guido, > > > > When passing multiple options on the commandline, this is (for me) > > > unexpected but easily fixed by passing all needed options into a single > > > --git-pbuilder-options, but when a gbp.conf file also uses > > > "pbuilder-options", these are also overriden by the commandline, which > > > is harder to fix (which needs copying the options from the config file > > > to the commandline). > > > > This in fact is intended since otherwise there'd be now way to override > > what's in gbp.conf (or rather in the several gbp.conf's parsed). > So you're suggesting to let multiple commandline options stack, but let > any commandline options override the config file option? That would seem > a bit confusing to me. Also, that would not allow using the commandline > to add additional options. > > My suggestion of adding a --git-append-pbuilder-option could solve both > usecases: > - you can use --git-pbuilder-options on the commandline to override all > previously set options, including in gbp.conf > - you can use --git-append-pbuilder-option to extend any previously set > options.
You would also need to define how option stack over the various gbp.conf files. I don't think we want to go down that road. -- Guido > At the same time, it would also keep backward compatibility as a bonus. > > (I used a singular, perhaps --git-append-pbuilder-options makes more > sense, though I guess that depends on whether argument splitting is > applied or not) > > > There are several levels of quiting going on, gbp itself does not do > > much but git-pbuilder and pbuilder do. > Good call. I did a bit of digging, turns out it is cowbuilder that > messes up, so this is entirely unrelated to this bug report. See > https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=898366 > > Gr. > > Matthijs
