Re: [gentoo-dev] [RFC/PATCH] A cleaner API for virtualx.eclass
On Tue, 12 Feb 2013 18:42:28 +0100 ""Paweł Hajdan, Jr."" wrote: > On 2/11/13 11:14 PM, Michał Górny wrote: > > My patches introduce a single wrapper with argv-as-parameter syntax. > > That is, the fore-mentioned example would look like: > > > > virtualx run_tests --foo > > Maybe we can just adapt Ubuntu's (I think) xvfb-run? More > standardization == profit. That's probably a good idea. However, I just stepped up here to replace the ugly API with something better. I think the actual eclass maintainers should have the final word here. -- Best regards, Michał Górny signature.asc Description: PGP signature
Re: [gentoo-dev] [RFC/PATCH] A cleaner API for virtualx.eclass
Le lundi 11 février 2013 à 23:48 +0100, Diego Elio Pettenò a écrit : > I'd say, "Go for it!" > > But on the other hand I wonder if it might make sense to have > something more generic, so that one only has to call something in a > way such as > > virtualx_setup > run_tests --foo > virtualx_cleanup This sounds nice but is imho but can be elaborated later on. -- Gilles Dartiguelongue Gentoo
Re: [gentoo-dev] [RFC/PATCH] A cleaner API for virtualx.eclass
On 2/11/13 11:14 PM, Michał Górny wrote: > My patches introduce a single wrapper with argv-as-parameter syntax. > That is, the fore-mentioned example would look like: > > virtualx run_tests --foo Maybe we can just adapt Ubuntu's (I think) xvfb-run? More standardization == profit. Thank you for working on this! Paweł signature.asc Description: OpenPGP digital signature
Re: [gentoo-dev] [RFC/PATCH] A cleaner API for virtualx.eclass
Am Montag, 11. Februar 2013, 23:14:38 schrieb Michał Górny: > > What are your thoughts? Same as Diego I like the general idea, but an even more generic framework might make sense. Say test dbus session, say setting up some test file structure, ... Oh, and one more thing... please before you commit anything talk to the kde guys... we inherit virtualx.eclass everywhere via kde4-*.eclass so we can easily use it (and many many kde packages need it for the tests). Changes in virtualx.eclass already broke many kde packages once... [IMHO I'd say it's perfectly fine if you determine the best way to go without specific kde regression testing, but before it goes in we'd like to have a chance to adapt our eclasses...] -- Andreas K. Huettel Gentoo Linux developer dilfri...@gentoo.org http://www.akhuettel.de/ signature.asc Description: This is a digitally signed message part.
Re: [gentoo-dev] [RFC/PATCH] A cleaner API for virtualx.eclass
I'd say, "Go for it!" But on the other hand I wonder if it might make sense to have something more generic, so that one only has to call something in a way such as virtualx_setup run_tests --foo virtualx_cleanup The reason why I'm wondering this is that we need some more "virtual environments" for testing purposes, really: so many packages need a D-Bus session (and I'd rather have them using a test session than a system one!), and at least in Ruby world we often need a database (sometimes more than one)... Diego Elio Pettenò — Flameeyes flamee...@flameeyes.eu — http://blog.flameeyes.eu/ On Mon, Feb 11, 2013 at 11:14 PM, Michał Górny wrote: > Hello, fellow developers, > > The current virtualx.eclass API is a bit insane. It seems a bit like > stacking of a few next APIs, mostly designed to quickly run 'make > check', then extended to general functions. > > For example running a function 'run_tests' with parameter '--foo' would > look like: > > VIRTUALX_COMMAND=run_tests virtualmake --foo > > which is really awful, considering that '--foo' is a parameter to > 'run_tests' and not virtualmake. > > My patches introduce a single wrapper with argv-as-parameter syntax. > That is, the fore-mentioned example would look like: > > virtualx run_tests --foo > > Depending on the maintainer decisions and your feedback, I believe that > even all the X* short-hand functions could be deprecated. They are a bit > shorter: > > Xemake check > > vs: > > virtualx emake check > > but I don't think that's much of a difference. > > What are your thoughts? > >
[gentoo-dev] [RFC/PATCH] A cleaner API for virtualx.eclass
Hello, fellow developers, The current virtualx.eclass API is a bit insane. It seems a bit like stacking of a few next APIs, mostly designed to quickly run 'make check', then extended to general functions. For example running a function 'run_tests' with parameter '--foo' would look like: VIRTUALX_COMMAND=run_tests virtualmake --foo which is really awful, considering that '--foo' is a parameter to 'run_tests' and not virtualmake. My patches introduce a single wrapper with argv-as-parameter syntax. That is, the fore-mentioned example would look like: virtualx run_tests --foo Depending on the maintainer decisions and your feedback, I believe that even all the X* short-hand functions could be deprecated. They are a bit shorter: Xemake check vs: virtualx emake check but I don't think that's much of a difference. What are your thoughts?