[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

testing boot floppies 3.0.18



Hi,

Here is a test report for the 3.0.18-2001-11-20 boot floppies.

So far, I have not filed any bug reports to the Debian BTS.  It is
not clear to me against which package any particular problem
should be filed against.

If you would like a bug report to be filed for any particular
problem, them please reply to this message, indicating which
package it should be filed against, and then I will file the bug
report.

Sorry for the cross-posting, please inform me as to which lists
are the appropriate ones to receive a test report such as this
one.


1. Install Method.

    o Off-line download of {rescue,root}.bin, drivers.tgz .

    o Off-line download of base archive using debootstrap, then
      created basedebs.tgz from the just downloaded archive.

    o Boot rescue.bin floppy, read drivers.tgz and basedebs.tgz
    from harddisk.

    o Install remainder of system via apt-get over a dialup ppp
    connection.

2. General Comments.

    o The install went pretty well, I was able to completely
    install the standard system, using dselect/apt over a dialup
    ppp connection.

    o Due to various problems encountered, I venture that a less
    experienced user would not have been able to complete the
    install successfully, and quite possibly would find that their
    prior boot configuration had been wiped out.

3. Problems Encountered.

    o The text based widgets are rendered incorrectly, eg, the
    shadow boxes are in the wrong place, and the scroll bar
    indicators consist of random hash (#) marks.  This happens
    during the main install phase, for all displayed screens,
    (prior to rebooting).  After rebooting, these same widgets are
    fine.

    o The directory selection screen for "install from harddisk"
    is preset with "/instmnt", this is wrong, it should be blank.

    o The lilo configuration process over-wrote the mbr on hda,
    even though I selected to install lilo into hda9.  This is a
    really nasty bug.  Under the potato install, lilo would have
    been installed into hda2, my extended dos partition, which is
    certainly better than over-writing the mbr, but I would prefer
    to have lilo installed directly into hda9 in any case.

    o The timezone configuration displays the choice that I had
    made, however, I am given no chance to cancel this choice and
    try again if I made a mistake or changed my mind.

    o The default lilo configuration makes it impossible to
    interactive with lilo at boot time.  This is really a poor
    choice for the default.  Instead, the default lilo
    configuration should contain these commands in the lilo.conf
    file,

        prompt
        timeout 300

    o The creation of the custom boot floppy can be speeded up,
    and made more reliable too, by using the "--superverify
    --verify-after" options together.

    In addition, the custom boot floppy takes about 90 seconds to
    be read when it is booted, compared to only 30 seconds for the
    rescue.bin boot floppy.  Since the rescue.bin image is used to
    obtain the linux kernel that is copied to the custom boot
    floppy, the boot characteristics, ie, how fast/safe it loads,
    should also be taken from the rescue.bin image, which I know
    will work, since I just booted from it.  In the case of boot
    from cd-rom, then just use the safe image as the default one.

    o After rebooting, I am asked if I want to startup the ppp
    connection, however, the pppconfig program is _not_ run before
    pon is executed.  A newbie user would be completely lost here,
    and under potato, pppconfig was run before pon was executed.

    o I have no pcmcia devices.  This was neither detected
    properly, nor was I queried about pcmcia, instead, there were
    various unsightly errors printed regarding pcmcia during the
    reboot.

    o My basedebs.tgz used base-config 1.29, which deleted all
    *.debs from the apt-cache, no questions asked, after the
    install had completed (and after having spent two hours
    downloading 30MB of *deb's).  This is very counter-intuitive,
    as I had already answered a prior question, in the negative,
    regarding removal of *deb's.


4. Summary.

    o From my perspective, there are just a few bugs to cleanup,
    otherwise, the install was smooth and easy.

Thanks,
-- 
Jeff Sheinberg  <jeffsh@erols.com>



Reply to: