Converting the 2.5 kernel to kbuild 2.5
I suppose a polite title for me would be "maintainer emiritus" of make config, make menuconfig, and make xconfig.
I'm in favor of abandoning the current tools because:
It's 3x maintenance to have 3 parsers for the same language.
It's difficult to do good syntax checking in scripts/Configure and menuconfig.
menuconfig in particular is too ugly to live.
A company which considers Linux its #1 enemy may own the copyright to "scripts/Configure". I don't know what kind of marketing or legal play they could make, but it would surely be hostile to Linux.
I'm in favor of CML2 in particular because:
ESR has designed a clean theory, which the configuration process really needs after ten years of ad hoc extensions.
ESR has done a lot of grunt work to turn a particular idea into a viable implementation. It's hard to get that work done.
As far as the Python issue goes, I believe that the kernel documentation just needs to state clearly what tools (and what versions) are needed to build a kernel. If other people prefer a C implementation, then CML2 (the language) is amenable to a C implementation, so they can write one.
As far as CML2 versus an mconfig-based solution, I am tilted towards CML2, as it is simply a better language. I would be happy with either choice if Linus made one of those choices. I would be unhappy if 2.6/3.0 continued to ship with Configure/menuconfig/xconfig.
Michael Elizabeth Chastain
<mailto:[email protected]>
"love without fear"
Quoted 4 lines · Michael Elizabeth Chastain
As far as CML2 versus an mconfig-based solution, I am tilted towards CML2, as it is simply a better language. I would be happy with either choice if Linus made one of those choices. I would be unhappy if 2.6/3.0 continued to ship with Configure/menuconfig/xconfig.
Indepenand of wether 2.6 will use CML1 or CML2 I hope it won't ship with the actual config tool. It's so much nicer to have mconfig compiled once in /usr/bin instead of compiling menuconfig all the time in the tree.
No to mention it's much easier to propagate bug fixes this way..
Christoph
Signature · 1 line
Of course it doesn't work. We've performed a software upgrade.
Quoted 6 lines · Christoph Hellwig
Quoted 4 lines · Michael Elizabeth Chastain
As far as CML2 versus an mconfig-based solution, I am tilted towards CML2, as it is simply a better language. I would be happy with either choice if Linus made one of those choices. I would be unhappy if 2.6/3.0 continued to ship with Configure/menuconfig/xconfig.
Quoted 6 lines
Indepenand of wether 2.6 will use CML1 or CML2 I hope it won't ship with the actual config tool. It's so much nicer to have mconfig compiled once in
/usr/bininstead of compiling menuconfig all the time in the tree.No to mention it's much easier to propagate bug fixes this way..
If the configure system is outside of the kernel, you have the possibility of requiring newer user-space utilities as a stable kernel changes over time...