Rockbox mail archiveSubject: Re: Bluechip forking
Re: Bluechip forking
From: sophana <jobarjo78_at_yahoo.fr>
Date: Tue, 08 Jun 2004 13:24:22 +0200
Björn Stenberg wrote:
>>the linux kernel has TONs of options (with xconfig...)
>>rockbox has a (very small) configure script. It should exploited better,
>>don't you think?
>More options means more builds, and more combinations in which to hunt bugs.
>And the fact is that most people (I'd guess >95%) don't know how to compile their own build, and so for the vast majority this work would be wasted.
I totally agree that options brings complexity. But it brings features also!
I don't think the option management would be wasted. It allows your mod
to evolve with rockbox.
making a web interface script for configuring and compiling rockbox from
cvs would be great.
there is actualy some code for maintaining all the archos hardware
versions. This same code could serve the additional optionnal rockbox
features? a specific make target would try to do a compile-regression
test with all option space.
maybe the firmware-linkable plugin feature would allow people to compile
their mod everytime with rockbox, but the user would choose dynamically
the features he wants by loading or not the plugin...
At work, we made a modified version of gmake and a set of rules which
does handle that kind of problems. Too bad it is not released.
All the great features contributed by lot of people make rockbox a great
piece of code.
The more feature there are the greater rockbox will be. (in my opinion)
>>The number of waiting patches is clearly high...
>Yes, but that is not helped much by making branches in the version control system. A branch is no different than a patch.
Received on 2004-06-08