Re: release update and branching
Dear Friends,
THe question I would like to have had answered, to wit, is the dual boot
disk geometry problem solved, has been. Thanks.
For the record, when using 2.6 to boot the cd, one sees that no matter
what, when checking the c,h,s before committing to the formatting, it
remains x,16,63, in rc2.
It rather sounds to me from the exchange herein as were the coders
either amused at the pain they will inflict upon the unsuspecting, or
willing to release an installer which they know full well will cost
hundreds of people thousands of hours simply because non-coders are
somehow less than human. This issue has been discussed since at least
June. At the moment, neither Ubuntu nor Debian can be installed on a
non LBA bios to dual boot with a pre-existing windows installation using
the ntloader.
As recently as yesterday, the issue came up in the DSLReports unix
forum, http://www.dslreports.com/forum/remark,12006676~mode=flat.
While no one respects the work here more than I, there is a lingering
prejudice that non-coders do not count. So, the d-i sort of works. I
have checked every week or so, getting as far as the partitioning, and,
when I see xxxxx.16.63, stop, as I know the result.
That is is silly that many notebook manufacturers use the
non-user-changeable Phoenix bios, or that windows uses the legacy
interrupts, &c, may be well true, but there has to be a solution before
the final, in my opinion. Obviously, the opinion of non-hackers is also
taken to be irrelevant, since the attitude has become more and more that
the only valid use of human intelligence is coding, and those who choose
to be lawyers, or musicians, or physicians who merely want to use debian
as they would windows must be somehow mentally deficient. We are not,
and actually understand the problem rather well.
Therefore, I will admit my sheer stupidity and ask if someone could
explain to the village idiot why one cannot simply prepare the free
space, have the d-i format the free space without touching the partition
table? Libranet used to do this, as does mepis. Suse provides a file
which one uploads during install to prevent the issue, anaconda
installers allow the passing of parameters hdx=c,h,s, and libranet is
not yet out with a 2.6 kernel, although the producers thereof told me
point blank that for a commercial distro such as theirs, the problem is
simply unacceptable. In other words, all one to simply install in a
pre-written fs. One builds an ext3 partition, or whatever, with a live
cd, leaves it empty but for the fs and table, then install cleanly onto it?
Why not make it possible simply to install the fs, then the OS, grub and
go? Or allow the parameters of the geometry to be passed to the kernel?
Both of these are fail safe. In theory, there is the parameter for the
"IBM" hard drive, but I have never been able to get it to work on on
non-IBM machine.
In other words, why make sport of what is a very serious issue, which
should in any ethical light completely prevent the issuance of the final
release? We are not talking about java not being point and click, or kde
printing manager not working on workgroup non-domain networks(it does
not, by the way, samba has decided that the simple peer-to-peer network
that sold so many copies of xppro to home users is somehow not safe for
a large corporation or school, so we, the small users should have to
learn real networking......). We are talking about people believing
themselves to have lost their other OS and most likely re-formatting,
thereby losing data. That is is not wise not to have tonnes of backup
is a given, but I should hope that kindness and compassion would take
precedence over intellectual acuity in such a serious matter.
That ubuntu states more or less clearly that Warty is not for those who
are not technical is at least an admission that the installer has a long
way to go before being something one can hand out to schools, teens,
friends, and so forth.
At some point, giving up and saying debian is only for those who really
want to learn about pcs rather than use them is sad, seeing how good the
installer is. In the end, it is true that arguing my daughter's school
does not teach Latin and Greek is petty, when seventy percent of the
world is illiterate, but it changes not the fact that she will grow up
without reading Plato. That few world wide users will encounter the
problem does not make it not release critical, since changing the
partition table is something an installer simply should not do, no
matter what.
Would one of you, therefore, be so kind as to tell me why securing
either the latest parted, using no parted, but formatting the free space
with something else, or allowing the c,h,s parameters to be passed to
the kernel is either so very difficult, or unimportant?
Or, would it not be possible, as a worst case scenario to publish
instructions as to how to use a file obtained with fdisk to re-write the
partition table after the 2.6 kernel install, in other words, use a live
cd to create a textfile somewhere(remember, floppy drives are no longer
de rigeur either), then use it to write the c,h,s indices back to where
ntloader could read them.
In other words, whatever solution one finds, the partition table must
clearly remain unaltered, or at least put something back in the barn
after the horse spillt the milk before we shut the door we should have
never opened before we burnt the barn down.
If the Ubuntu folks, Watson and friends solve the issue, it would surely
be a matter of extreme urgency to use that solution in the final.
"and let users decide whether to install with 2.6 or not and risk the
problem" is not solution at all, as those users who understand the
problem will not go forward with the partitioning, and use expert 2.4 to
boot, then install 2.6, as I did, and those who do not will be in a
blind panic.
To allow this is inhumane, therefore, a netinst iso with the latest
parted needs be tested as soon as possible. If someone can create on cd
with all other things the same, but the latest parted, I shall be
happy to test it, as I know enough not to go past the danger point if
the read is not correct.
For your information:
yazdzik@deblap1:~$ su
Password:
deblap1:/home/yazdzik# parted
GNU Parted 1.6.11 with HFS shrink patch 12.5
Copyright (C) 1998 - 2004 Free Software Foundation, Inc.
This program is free software, covered by the GNU General Public License.
This program is distributed in the hope that it will be useful, but
WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or
FITNESS FOR A
PARTICULAR PURPOSE. See the GNU General Public License for more details.
Using /dev/hda
Warning: Unable to align partition properly. This probably means that
another
partitioning tool generated an incorrect partition table, because it
didn't have
the correct BIOS geometry. It is safe to ignore,but ignoring may cause
(fixable) problems with some boot loaders.
Ignore/Cancel? cancel
Information: The operating system thinks the geometry on /dev/hda is
116280/16/63. Therefore, cylinder 1024 ends at 503.999M.
(parted) q
Information: Don't forget to update /etc/fstab, if necessary.
deblap1:/home/yazdzik# fdisk -l /dev/hda
Disk /dev/hda: 60.0 GB, 60011642880 bytes
255 heads, 63 sectors/track, 7296 cylinders
Units = cylinders of 16065 * 512 = 8225280 bytes
Device Boot Start End Blocks Id System
/dev/hda1 * 1 3697 29696121 7 HPFS/NTFS
/dev/hda2 3698 7184 28009327+ 83 Linux
/dev/hda3 7185 7296 899640 82 Linux swap
deblap1:/home/yazdzik#
Obviously, using 2.6 as the installation kernel would result in the
problem. Since I cannot change the bios read on my laptop, windows will
use the 7296,255,63 whenever installed. There is no other possibility.
Thereofre I need the partitioner to leave the table alone.
At the moment, neither Ubuntu nor debian could be installed without a
major workaround, in debian, the one mentioned as installing in 2.4
expert then choosing 2.6 at the prompt, in ubuntu, one could have to
copy a file from fdisk, then re-write the table after the install.(This
is a genuine experts-only undertaking.)
I hope someone will at least re-examine the severity of the issue, as if
parted were the only major change, it might indeed be worth it to be the
only distro with a 2.6 kernel and an installer that works for the
inexperienced. It would indeed be at the least less than courteous to
release a final without a sever warning to peruse and understand the
significance of the c,h,s entries in the expert mode, rather than the
usual, "that's your funeral" that dissuades few from heroism, as that
disclaimer has too often cried wolf.
Many thanks for the immense, heroic, unfathomable work put into this.
Take not my directness for lake of gratitude, but as a statement of
belief that the installer should have a far wider, and far safer,
audience. It, and the development team, deserve that.
Warm regards,
Martin
Sven Luther wrote:
On Thu, Dec 02, 2004 at 03:19:47PM -0500, Joey Hess wrote:
Sven Luther wrote:
Well, i don't really anticipate doing loads of parted 1.6.11 uploads anyway,
but i guess this is moot, since it seems there is a rough consensus among the
RMs to get the experimental parted in sarge. Or at least try to do so.
Unless you really like to reply to all the future "debian ate my windows
install" bug reports :)
I think I would prefer to do so, or properly document the issue and let
users decide whether to install with 2.6 or not and risk the problem
what about all those that have hardware which is not supported by 2.4 ?
(which doesn't seem to happen very often), than make a rc3 release. I'm
done with major d-i releases for sarge. If someone else wants to
coordinate another d-i release for sarge with lots of large changes and
wants to include a new parted in it, then maybe.
Let's let the ubuntu guys test it, and then we will see :)
Seriously, it would be nice to have a experimental version of the neinst iso
or whatever with the partman/parted changes, so we can ask users with problems
to test that.
That said, if we had done the change in august, then we would not have this
problem anymore now, so the whole discussion comes down to if we will release
this year yet (highly unlikely), or 6 or more onth from now.
Friendly,
Sven Luther
Reply to: