What is LFS anyway?
jeremy at jutley.org
Sun May 9 20:17:43 PDT 2004
Bill's LFS Login wrote:
>On Sun, 9 May 2004, Bruce Dubbs wrote:
>>Alexander E. Patrakov wrote:
>>I would be in favor of expanding Chapter 9 to tell the user *how* the
>>new system can be customized for individual needs. This includes using
>>chroot from the host system until those packages the user needs are
>>installed. Suggested packages to install early for a desktop system
>>would include communication (dhcp client, wget, etc) and X+Window manager.
>With the availability of VT's, I consider X unneeded for a *base*
>platform. Just include gpm, most of BLFS Chapter 3 stuff, and a
>*minimal* browser. That plus network connectivity seems enough to me.
>Think about it: what's the *big* difference between doing X+* in the end
>of Chapter 9 and doing it as the (user chosen) early effort of BLFS? A
>lot in terms of book size, because of dependencies, and little
>conceptually. If the user has the foresight to download BLFS and a
>browser, we can even eliminate the net connectivity rqmt. If the user is
>comfy with CLI, only convenience is lost if no gpm (of course this is
>easily overcome with a text version of books and vi). Wget? A bash-only
>solution was posted many moons ago (inelegant to read, but so what?)
>My feeling is that if LFS taught what it intended, had a *base* platform
>from which a user who has *learned* can accomplish his/her goals, LFS is
>done. I even feel that if they must reboot into their old host, that's
>OK too. If the book wants to progress beyond that, also OK but no one
>would call it a base platform, by the time prerequisites and X are
Forcing the user to reboot back into their host constantly after
finishing the LFS is just wrong, IMHO.
Especially when adding 3 packages will give about 99% of people the
ability to browse in their new LFS
successfully (lynx, dhcpcd, ppp). Quite frankly, very few people have
static IP ethernet connections,
needing either DHCP or PPP(oE). In fact, I'd probably say the only
people who *DO* have static IP
access are the ones who'd already know *HOW* to set up networking,
because probably anyone with
static IP access has their own linux router already.
>Creeping "bloatitis" is difficult to resist. Division of labor is
>important. Focus is easily lost.
I don't call making LFS functional after the book is creeping bloatitis
- I consider it doing the right thing
for the first time LFSer - the experienced LFSer won't need the help,
but the first timer will. I think the
biggest problem is those of us here on -dev have forgotten what it was
like the first time they booted into
that shiny new LFS, then tried to install further stuff to make it a
functional system., and found they couldn't
get anything because no functionality was there to do so. It's better
now that we have an ftp client (provided
so kindly by inetutils), but I think we could easily do better, without
"bloating" up the LFS install.
><rant not related to you at all>
>Especially when a crew that wants to be bleeding edge is holding sway
>and denigrating valid concerns with whines about resisting the future.
>A low form of argument if ever I heard one, but maybe they are
>incapable of better. Or respect none but their own opinions and
></rant not related to you at all>
This smells an awful lot like a personal attack against quite a few
people who really want to see LFS
move forward. Lets face it...2.6 is here, has been for months, and so
has NPTL - isn't it about time we
were using it? Udev - no it's not essential now..but soon will be, when
the kernel guys start randomizing
device numbers, so we might as well get it going - it also solves the
long-standing issue with device creation
(too many or too few). Hotplug is just simply a good thing to have, and
is the preferred method of dealing with
removable devices, such as USB or Firewire devices.
Just a few weeks ago, people were complaining because there was no
development, now there's development again
and people are STILL complaining. Make up your mind guys, please?
More information about the lfs-dev