Re: Some reflections on the future of partman [was: Partman]
Anton Zinoviev schrieb: 3. See how this problem is solved by SuSE - YaST supports both text and graphical installs, both have very nice look and as far as I know there is almost no duplicate code (due to text+graphics). Not everything is good, that looks pretty. Some weeks ago I tried to install Suse 9.0 on my test machine, ordinary Celeron 600 with a 106 GB IDE-HDD. hda1 - hda8 are used by other installations. Suse partitioner was unable to create a hda9 und did not recognize the full 160 GB disk size. Thus I installed Suse under VMware with sarge as host, which was straight ahead. There it should live, captured on a single virtual disk. Other situation, other box. Knoppix 3.4 hd-install uses now a graphical partitioner, gtk-parted AFAIR, instead of good old cfdisk. hda1, hda2 and hda5 were used by others. I wanted Knoppix to go into hda6, but creating a logical partition was impossible. I am not happy with this Linux partitioner confusion. I my dreams I see a partitioner with full, bugfree functionality and exchangeable user interfaces on top, but I am too unskilled to develop it. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Expect for d-I tests
Petter Reinholdtsen schrieb: I suspect it will be easier with bochs, as qemu is missing a feature. bochs can use the 'term' console driver, and then use normal stdin and stdout as the console. This console would be easy to control using expect. Sounds interesting for test automation. Since a few days I have VMware, which is nice for playing around with installers, but allows only screenshots at early installer phases. What's about UML instead of bochs? Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#273215: [i386] [USB] [errors] installation report
in Debian since years. Possible solution(s): - redefine the requirements of d-i - redesign the navigation: make all screens/functions hierarchically ordered, step back etc. - take into account, that steps can fail, and exception handling should be possible at UI-level 10) Does not install UNICODE or ask for it -- Problem: BUG Severity: IMPORTANT (NORMAL?) Description: Got en_US.ISO-8859-1 installed and configured. d-i did not ask for another option. UTF-8 should be the default, other locales optional. 11) No pager in base install Problem: WHISH Severity: NORMAL Description: Cursor keys do not work in man. 12) Important packages not (un)selectable/choosable/included in tasksel --- Problem: BUG (it's really more than an important whish) Severity: IMPORTANT Problem: Some packages, which nearly everybody wants to install, are not installed - like ssh, less. On the other hand unbelievable amounts of packages are installed, not meeting the needs of everyone. E.g. I never use Open Office, but I like KDE. Possible solution(s): - redefine the tasks to a finer granulation - provide a usable (=Usability) package selector 13) X misconfigured --- Problem: BUG Severity: IMPORTANT Description: See relevant part of log: (II) APM: driver for the Alliance chipsets: AP6422, AT24, AT3D (II) Primary Device is: PCI 01:00:0 (EE) No devices detected. Fatal server error: no screens found Ok, looked into my old conf and did the same, what Knoppix does: fall back to vesa, and Buttons "5". Possible solution(s): - learn something from Knoppix or Suse 14) GRUB did not install all previous OS Problem: BUG Severity: IMPORTANT Description: There were 2 bootable OS installed, but only one of them is recognized. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Bug#279307: debian-installer: shortcut for language/country/keymap is useful
Christian Perrier wrote: The keyboard layout will, however, remain preseedable distinctly as the code for handling locales and keyboard layouts is very separate codemaking things too much interdependent is probably not a good long-term idea... That's the point. Assuming much seems to be convenient for mainstream, but gets inconvenient in case of special needs. E.g. I want to have English as language for installation, administration, messages and man, as this has advantages to find similar problems or describe them in newsgroups, forums and bug trackers. Keyboard is German, location is Austria (but can change in case of a notebook), encoding should be UTF-8. Developers should always keep in mind, that different users can exist on one system, that one user can have different roles, wants to use more than one language at the same time etc. Thus locales should always be choosable and reconfigurable in a convenient way without reading tons of manuals. Assuming interdependencies, and using inheritance to get senseful defaults is o.k. But the user should always have the chance to override the defaults, or to change to expert level or even deeper expert levels. BTW (not a problem of d-i, but a problem of installation) we can learn from the different behaviour of some packages. AFAIK Gnome installs English, if the base-installation is Englisch. KDE asks the user after the first login, which language etc. he wants to have, but some language specific packages need to be installed manually (e.g. spell checkers). Open Office in the other extreme installs with many languages, causing huge downloads of unneeded files. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Bug#279307: debian-installer: shortcut for language/country/keymap is useful
Christian Perrier wrote: Quoting Helmut Wollmersdorfer ([EMAIL PROTECTED]): Well, I do not see any problem here..:-) All this is technically possible with the current d-i...except defaulting to a UTF-8 environment (that will be possible as soon as localechooser replaces lang+countrychooser) I wonder about this, because UTF-8 is the only general solution against the historical character encoding babylon. I do not agree about only using English for system administration but this is a personal feeling which is for sure not enforced anywhere in d-i ACK. I absolutely don't know where you got the idea that all this is not possible..:-) [...] For this, you need to either use the expert install...or "dpkg-reconfigure locales" when the installation is over. Thus locales should always be choosable and reconfigurable in a convenient way without reading tons of manuals. This is exactly how d-i behaves since the beginning..:-) You mean that d-i fullfilles _all_ of the criteria of my above sentence: 1) always 2) choosable and reconfigurable 3) in a convenient way 4) without reading tons of manuals No. Using something like "dpkg-reconfigure locales" needs a newbie, or even an experienced linux user but new to debian, to know the exact spelling, or reading manuals, or writing down by hand all messages of d-i. Hopefully sid will a have a more general, menu driven solution, where _each_ screen has back, up, help, skip and expert choice. I don't want to be misunderstood. Debian developers made a great job to make a large scale of configurations, architectures and exotic combinations _technically_ installable. But there is still a great lack in usability. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#364638: Invalid characters in short user name need reboot
[sorry for maybe breaking the thread reference, but gmane is not working for me at the moment] Christian Perrier wrote: [...] Does not handle non ASCII input properly I'm first reassigning this to cdebconf even though I'm not sure this really belongs to that package. With my little knowledge of d-i source I assume it somewhere in SVN/packages/cdebconf/src/modules/frontend as the gtk-plugin does not have this behaviour. Indeed, at any point in the installer, all non ASCII input in select questions is unproperly handled. You mean 'string questions'? In 'select questions' the non ASCII keys do not work too, but do not have a bad side-effect. E.g. in one of the last dialogs of the partitioner there is a choice in the German version beginning with a non ASCII 'Änderungen auf Platte schreiben' and jumping to this choice does not work by clicking the key 'ä' or 'Ä'. HTH Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#365308: Example of what happens when entering non ASCII characters in a select field
Christian Perrier wrote:
Attached is a snapshot of what's displayed when I enter "Christian
Périer" as user's full name in user-setup (I used full name rather
than login name so that I can see what happens later when the user is
created and whether the right encoding is used in what's passed ao useradd)).
Please note that "" is the code for "é" in ISO-8859-1. So it
barely seems that the keyboard input is interpreted as encoded in ISO
and not UTF-8.
As I develop a test robot and OCR (Optical Character Recognition) is
ready, I attached screen shots in plain text.
They contain the keystrokes of the German keyboard in the order 1st row
left to right , next combined with , next combined with
, next row ...
Helmut Wollmersdorfer
Please press Enter to activate this console.
BusyBox v1.01 (Debian 1:1.01-4) Built-in shell (ash)
Enter ´help´ for a list of built-in commands.
~ # ^1234567890ô °!"§$%&/()=?` ²�{[]}\ qwertzuiopü+ QWERTZUIOPÃ*
@wertzuiop~ as
dfghjklöä# ASDFGHJKLÃô asdfghjkl YXCVBNM;:_ |yxcvbnµ �
string is
GETTING HELP F9
If you can´t install Debian, don´t despair! The Debian team is ready to
help you! We are especially interested in hearing about installation
problems, because in general they don´t happen to only one person.
We´ve either already heard about your particular problem and can dispense a
quick fix, or we would like to hear about it and work through it with you,
and the next user who comes up with the same problem will profit from your
experience!
See the Installation Manual or the FAQ for more information; both
documents are a
Bug#364638: Bug#365308: Example of what happens when entering non ASCII characters in a select field
Helmut Wollmersdorfer wrote: ready, I attached screen shots in plain text. Sorry, I didn't switch the header to UTF-8. The attached files of the previous post are best viewed with a UTF-8 capable editor - e.g. kedit works nice. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Proposal for integration of the graphical installer
Stefano Canepa wrote: Hyphen is a problem on Italian keyboard, too. So using only letters could be better for us. It's a problem on nearly all European keyboards. Only '!$,.', 0-9, a-zA-Z are in the same position. And there are also letters with different positions, e.g. 'zy' on German keyboards. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: versions of debian-installer
Geert Stappers wrote: The full URL contains the build date. Hmm ... if I go to http://www.debian.org/devel/debian-installer/ then 'To install Debian testing ...' gives e.g. http://cdimage.debian.org/cdimage/etch_di_beta2/i386/iso-cd/debian-testing-i386-netinst.iso 'If you'd like something newer ...' http://cdimage.debian.org/cdimage/daily-builds/daily/arch-latest/i386/iso-cd/debian-testing-i386-netinst.iso That's the reason, why I always download into a directory d-i-iso/ and create a textfile with download-URL and build-date (from boot-screen) in it. The installed system and the ramdisk during install contains the file /etc/lsb-release which has build date. Thx for the hint. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Bug#379526: papersize is a4 for US install - should be letter
Christian Perrier wrote: So, if you actually want a paper size adapted to your country, I'd recommend to set the locale properly. ACK. This can be a nice _default_. But keep in mind that the papersize depends only on the piece of paper which I put into (one of) my printer(s). Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: [RFC] Detecting type of CD image and using that to make apt-setup smarter
Goswin von Brederlow wrote: Ok, slightly different point: When etch R1 comes out people will still install R0 and want the updates from the mirror. Which can have a very bad effect: In October 2005 I installed a sarge-R0-netinst.iso using a nearby mirror for additional packages. But the security mirror was also configured, and the updates needed endlessly (overnight, >10 hours), because all/many users updated from the one and only security 'bottleneck' mirror. If it will still be the case with etch, that - there is only one (slow) security mirror AND - important updates are held back until R1 it would be nice, to ask the user, if he really wants the security updates 'now', or 'later'. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Debian and others installation
Roseland, Winton wrote: After service pack 6A, NT started to lock up on it, so I went to Win2K but after a service pack I started getting lock-ups again where it appeared that the IDE controller did not respond. Linux has worked well, as Knoppix will run for months before I reboot. 1) I have a HP PSC 1315 printer attached to a USB 2.0 PCI card, and all recent Debian based versions died during the install as long as the printer was plugged in. A message came up about initializing or checking the USB and that was the end. There were no errors: it just never did anything else. Did you consider a problem with your hardware? My workstation often (more than once a day) got freezes under Win XP, but very seldom (once a month) under Debian Sarge. This was mostly after opening a large file. A boot into memtest showed my very quickly, that the memory is corrupt. After changing I never had freezes. There can be a lot of other reasons like damaged capacitors on the mainboard, bugs in the chipset drivers, a weak power supply etc. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: [RFR] Proposal for installs without network connection
Bill Allombert wrote: On Tue, Apr 04, 2006 at 10:19:24PM +0200, Frans Pop wrote: With Beta 2 we have lost the option to (easily) install without using a network connection. There have been several comments about this. Please keep in mind that there is a distinction between having no network connection and no Internet connection. This distinction is not important, as somebody can install from a mirror in the LAN. IMHO Frans is addressing the problem, where somebody _wants_ or _needs_ to install from CDROM only. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: [RFR] Proposal for installs without network connection
Frans Pop wrote: On Wednesday 05 April 2006 00:06, Joey Hess wrote: -- but it detected it was a full CD and just asked the user to make sure they didn't want to use a mirror too. Your proposal only approaches that behavior if a user chooses not to configure the network, which most users will not do. Anyone who configures the network will still have to go through choose-mirror. But now has the added option to select "don't use", which does approach the old behavior. But 'no_net' is confusing for this purpose. If somebody wants to use CDROM only such a option should have the name 'cdrom_only'. Whether or not to default to that option for full CD installs is open for discussion. The answer is very easy: If somebody downloads/burns/uses the huge full CDROM, we can assume that he wants to install from CDROM. Reasons can be standalone, slow, or expensive internet connections etc. Effectively this implements the old "do you want a mirror too" question. IMHO it is very inpolite behaviour to configure a mirror without knowledge/asking the user. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
success - CD-boot with floppy-preseed
Because it is not documented: On the boot prompt enter after '... boot:' install preseed/early_command=mountfloppy preseed/file=/floppy/flpre.cfg Maybe this is useful for somebody, who cannot create a custom CD or use http. I tested this with VMware Workstation 5.5, a daily debian-testing-i386-netinst.iso dated 2006-03-29, and a floppy.img. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: Bug#363406: acknowledged by developer (Re: Bug#363406: installation-report: X does not work: glx not available for ATI Radeons)
Frans Pop wrote: The severities critical, grave and serious are intended to be used _only_ for issues that are Release Critical. First is the definition of severities on http://www.debian.org/Bugs/Developer#severities, 'Release Critical' is another problem. That is: if they are not fixed, the next release of Debian should be postponed. Yes, makes sense. Do you really think this qualifies? Yes. Please... ... imagine the release of a distribution which ends in black screens for wide spread graphic cards. The configurator could fall back to 'vesa', as AFAIR Sarge did on my notebook with an ATI 9700 (or I did it by manually editing?). No, it is not unusable. The system works fine, it is just one loasy program on it that does not work because it is not configured correctly. Impact for the user: high (depending on his proficiency with Linux); impact for the system: null. It makes unrelated software on the system break, (critical) OR at a minimum It has a major effect on the usability of a package, without rendering it completely unusable to everyone. (important) IMHO it's a bug. But maybe against the wrong package. IMHO you should direct it to the correct package. Closing 'unsolved' is not a solution. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Re: success - CD-boot with floppy-preseed
Joey Hess wrote: The mountfloppy thing is not necessary; preseed/file automatically mounts /floppy if set to use a file from there. Thx, you are right (just retested). There must have been other reasons for the error messages I got. FYI, preseed/early_command is only run *after* the preseed file is loaded. After the _first_ preseed file or after _all_ preseed files? Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#364631: German keyboard not fully supported in the console (ash) of the installer
Severity: minor Package: installation-reports Boot method: CD-image in VMware Image version: http://cdimage.debian.org/cdimage/daily-builds/daily/arch-latest/i386/iso-cd/debian-testing-i386-netinst.iso 'built on 20060409' Download date: 20060420 Date: 20060424 Machine: VMware Workstation 5.5.1 under Debian/unstable Test sequence: - boot without parameters (=default) - language: German - keymap: German - country: other -> Austria - after automatic DHCP and prompt for hostname, + to console, - type in all keys and watch behaviour Expected result: all characters engraved on the standard German keyboard should work Actual result: some are not working = display wrong characters +e displays e instead of EURO SIGN (U+20AC) +3 displays small filled quad instead of ³ SUPERSCRIPT THREE (U+00B3) Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#364638: Invalid characters in short user name need reboot
Severity: important Package: installation-reports Boot method: CD-image in VMware Image version: http://cdimage.debian.org/cdimage/daily-builds/daily/arch-latest/i386/iso-cd/debian-testing-i386-netinst.iso 'built on 20060409' Download date: 20060420 Date: 20060424 Machine: VMware Workstation 5.5.1 under Debian/unstable Test sequence: - boot without parameters (=default) - language: German - keymap: German - etc. - at the prompt for the user account name type in a non-ASCII character like 'ä' LATIN SMALL LETTER A WITH DIARESIS into the edit field Expected behaviour: --- 1) it is not possible to enter invalid characters OR 2) let the user correct the string after an error message Actual behaviour: - - The edit field displays the string '' instead of 'ä', - the yellow background signalling the cursor position hangs on the '<' of '', - or is not working, - after an error message and managing going back to the question, the field again is initialized with the defect string -> no chance for correction, reboot needed Characters not working are: ß°§üÜöäÖĵ All other invalid characters (outside [a-z][a-z0-9]+) of the German keyboard do not damage the field editor. In case of other strings, e.g. hostname, it is possible to go to the main installer menu, retry, and a nice and sane edit field is displayed. BTW: g-i has better handling in the same situation. PS: maybe you want raise the severity higher. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
Bug#364631: German keyboard not fully supported in the console (ash) of the installer
Frans Pop wrote: Could you please send the output of: # grep "kbd-chooser\[" /var/log/installer/syslog As '/var/log/installer/syslog' does not exist, I did # grep 'kbd-chooser\[' /var/log/syslog Apr 25 13:25:02 kbd-chooser[3755]: DEBUG: Mounting usbdevfs to look for kbd Apr 25 13:25:02 kbd-chooser[3755]: INFO: keyboard type at: present: true: Apr 25 13:25:02 kbd-chooser[3755]: INFO: Setting debian-installer/serial-console to false Apr 25 13:25:02 kbd-chooser[3755]: INFO: Setting debian-installer/uml-console to false Apr 25 13:25:02 kbd-chooser[3755]: INFO: kbd-chooser: arch at selected Apr 25 13:25:11 kbd-chooser[3755]: INFO: choose_keymap: keymap = de-latin1-nodeadkeys Apr 25 13:25:11 kbd-chooser[3755]: INFO: kbd_chooser: setting keymap de-latin1-nodeadkeys IMHO this looks o.k. Helmut Wollmersdorfer -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
