[Hurdextras-hackers] Who needs a clue-by-four

[James A Morrison]

 Hey,

  Anyone remember that guy with the tarball?  Well, it's a weird indication
that some people would rather write than read.  Anyway, here is the actual
text that was sent to help-hurd.  

  Here is the breif synopsis:  I don't know any terms to use.  I think 
translators are great, but I still don't get passive translators.  I am a
newbie, HEAR ME ROAR.  GNU/Hurd acts sort of like GNU/Linux.

Jim

INTRODUCTION:

In this feeback report I will try to present a few simple problems I have 
encountered
in my very first contact with Hurd.Perhaps is this presentation of no value.On 
the other
hand,maybe you can find a few things which are interesting,helpful,of 
constructive value
and contributing to the work and development of GnuMachHurd.Unfortunately,I 
have no know-
ledge of even basic principles in programming,debugging and so forth.Therefore 
it is some-
what embarrassing for me even to try something for the GnuMachHurd.One of the 
problems I
encounter in this respect,is that the article most probably will lack 
information of screen
messages and other checkouts important for the serious working people of Hurd.

Besides this,I do not need any reply-messages concerning this report;do not 
bother your-
selves except that: 1) you feel you have to 2) you are used to replies all the 
time and 
therefore cannot stop it 3) the content of the article is of value or interest 
4) you feel
you have to be polite 5) other reasons that I cannot think of.

###########################################################################################

THE LOGINS:

Command:        login > sigbj
Result:         login prompt (bash):sigbj:command not found

Command:        login > root
Result:         login prompt (bash):root:command not found

Command:        login > ql sigbj
Result:         Password:
Result:         address@hidden:~$
Command:        ll
Result:         #here the aliases work, uncommented and made in '~$ .bashrc'

Command:        login > su root
Result:         bash:mesg:command not found      #no password set here for 
simplicity

Command:        login > su sigbj
Command:        ll
Result:         command not found        #here the aliases do not work
Command:        su root
                su sigbj
Result:         #now the aliases work

Comment:        It seems as if the 'login > ql' make the shell read the 
.bashrc. Otherwise
                login > su does not do this immediately.The login > USER does 
not work for
                some reason.

###########################################################################################

THE TRANSLATORS:


Prompt:         address@hidden:~$ dd if=/dev/zero of=dummy.fs bs=1024k count=8
Prompt:         hurd:/home/sigbj # mke2fs dummy.fs
Prompt:         address@hidden:~$/personlig  dd if=/dev/zero of=dummy.fs 
bs=1024k count=8
Prompt:         hurd:/home/sigbj/personlig # mke2fs dummy.fs

Comment:        I make 2 filesystems in 1)my homedirectory 2)an additional 
directory

                login > ql sigbj
Prompt:         address@hidden:~/personlig$   # I am in my secondary user 
directory
Command:        settrans -ac dummy /hurd/ext2fs dummy.fs
Command:        ps -x
Result:         87 -So 0:00.01 /hurd/ext2fs dummy.fs
Command:        ls -li
Result:         2   drwxr-xr-x   3 root   root   1024   Nov 4 17:45   dummy
Command:        settrans -g dummy    #short delay after 'enter'.
Command:        ls -li
Result:         467   -rw-r--r-- 1 sigbj   users   0      Nov 4 17:49   dummy

Comment:        This is quite fascinating to me: to see that a directory is 
created by
                the start of the translator and the directory turning into a 
file by
                the termination of the translator.This is GnuHurd and nothing 
else!

Prompt:         address@hidden:~/personlig$   # I am still in my secondary user 
directory.
Command:        settrans -p dummy /hurd/ext2fs dummy.fs
Command:        ps -x
Result:         #no server process yet
Command:        ls -li
Result:         2   drwxr-xr-x   3 root   root   1024   Nov 4    dummy
Command:        ps -x
Result:         130 -So 0:00.01 /hurd/ext2fs dummy.fs
Command:        settrans -g dummy  #short delay
Command:        ls -li
Result:         467   -rw-r--r--  1 sigbj   users   0   Nov 4   dummy

Comment:        I just try to show here that the settrans command works like it 
should.
                I also created a directory mnt/.Settrans worked like it should 
and 
                the lost+found directory appears in mnt/ in the same fashion as 
with dummy.


Command:        settrans -a /mnt /hurd/ext2fs dummy.fs
Result:         settrans: /mnt: Operation not permitted
Comment:        This is just what is expected trying out on directories where I 
have no 
                admission.
Command:        ps -x
Result:         57 -So 0:00.00 /hurd/ext2fs dummy.fs
Comment:        Watch it! Something is going on!
Command:        settrans -a /mnt /hurd/ext2fs dummy.fs  #I set the command a 
second time!
Result:         ext2fs: dummy.fs: warning:filesystem not unmounted 
cleanly.Please fsck.
                ext2fs: dummy.fs: warning:mounted read-only.must use fsysopts 
--writable

Comment:        This I find interesting.Because now this message follows me 
with every
                settrans attempt.I can try 'kill 57'.The process disappear,but 
it does
                not change the warnings.The only way to get rid of it is to 
reboot the
                machine.It is always on the second settrans this happens.I have 
a certain
                feeling why this is so,but should this be considered a bug?
                I am not sure but looking at the time interval of PID 57 it is 
set to
                0:00.00 and not to 0:00.01 which I find in most cases with 
succeeded
                settrans -a actions.Am I touching a point? A process is 
half-way started,but
                suddenly the system finds out that permission lacks.This stops 
the process
                half-way,but "too late":the warnings are already set.I could 
assume that
                the warnings were written on a thereto designed block in the 
dummy file-
                system,but why should a reboot erase this? So I really do not 
believe that.
                Is this so,and is it a bug?

Prompt:         address@hidden:~$   # watch it now! I am in my primary user 
directory.
Command:        settrans -ac dummy /hurd/ext2fs dummy.fs
Result:         #To be short: active translator works like it should.I do not 
need to go
                #into details here.There are no other messages.
Command:        settrans -g dummy       #OK
Command:        settrans -p dummy /hurd/ext2fs dummy.fs
Command:        ls  or  ls -l or ls -li
Result:         Translator died. # there are no more messages,but on ls or ls 
-l(i) the
                file 'dummy' disappears! It is not any longer there!NOTE THIS!
                Also note that upon 'login > su sigbj' where the 'ls' do not 
work(see
                the writings below on next topic),the dummy is there,however 
when I
                start the translator with 'ls -l(i)' -which works - the file 
'dummy' by
                the translator turned into a root-owned directory disappears 
together with
                the appearence of the message 'Translator died'.
Comment:        What so ever I do:IT IS IMPOSSIBLE TO START A PASSIVE 
TRANSLATOR IN MY
                PRIMARY HOME DIRECTORY AS USER OR EVEN AS ROOT!!This is 
difficult to under-
                stand.There might be an answer among the mainbrains.I have not 
the slightest
                idea why.ALSO NOTE THAT A FEATURE IN THIS RESPECT IS THAT THE 
MOUNT POINT
                FOR THE FILESYSTEM (in this instance 'dummy') CANNOT BE 
DETECTED BY THE
                COMMAND 'ls' IN ITS DIFFERENT FASHIONS WHEN THE TRANSLATOR IS 
TRIED STARTED.
                HOWEVER 'dummy' COMES BACK WITH 'settrans -g dummy' AND THEN 
'ls -l'.


##############################################################################################

LOGIN > SU USER, USER .BASHRC, AND PASSIVE TRANSLATOR START-UPS


Login:          login > su sigbj        #(and Password also of course)
Prompt:         address@hidden:~/personlig$
Command:        settrans -p dummy /hurd/ext2fs dummy.fs
Command:        ps x
Result:         #no server process as expected
Command:        ls
Command:        ps x
Result:         #still no server process!
Command:        ls -l   or ls -li
Result:         #server process starts as usual!!
Comment:        This means that on the command 'ls' even if repeated does not 
start the server
                process of the /hurd/ext2fs. Only when I use ls -l or 
similar.The same
                thing I have found if I as 'su root' start the translator on 
/mnt mounting the
                /home/sigbj/personlig/dummy.fs: the ls does not work.Remember I 
test the pro-
                cess with ps -x.My own question is if my [.bashrc] or shell 
started in my home
                directory by 'login > ql sigbj' has anything do to with this. 
login > su sigbj
                does not read my .bashrc as I wrote in the above,but the login 
> su sigbj
                reads ls -li and ls of course,only not aliases like 'll' and 
'..'.
                Is all this just plain nonsens from me or has it value? That's 
my question to
                myself.My user (sigbj) .bashrc is uncommented at the following 
aliases:

                        eval 'dircolors -b'
                        alias ls='ls --color=auto'
                        alias ll='ls -l'
                        alias l='ls -CF'
                        alias ..='cd ..'


##############################################################################################

LOGIN > SU ROOT, ROOT .BASHRC, AND PASSIVE TRANSLATOR START-UPS


Login:          login > su root
Command:        cd /
Command:        settrans -p /mnt /hurd/ext2fs /home/sigbj/personlig/dummy.fs
Command:        ps -x #no process yet like expected
Command:        ls
Command:        ps -x #still no process
Command:        ls -li
Command:        ps -x #short delay and process show up!
Comment:        In this case I let roots .bashrc stay commented with '#'.If I 
uncomment
                the line no.1  [ export LS_OPTIONS='--color=auto ]   reboot and 
try the same
                procedure the same result show up.If I uncomment the next line 
no.2
                 [  eval 'dircolors' ] still the same result.If I uncomment 
line no.3
                [alias ls='ls $OPTIONS'] and reboots,then the 'ls' displays the 
root directory
                tree in color,get the delay during the start of the 
/hurd/ext2fs filesystem
                server and the ps -x displays the corresponding started process 
with its PID.
                If I now recomment with '#' lines no.1 and no.2 in roots 
.bashrc,reboot the 
                machine and try the procedure, the 'ls' will show no 
color-dir-tree and 
                start no process.I have to again set 'ls -l' or 'ls -li' to 
start the server.
                I DO NOT UNDERSTAND THE MECHANISM BEHIND THIS.But there are of 
course several
                possibilities: 1) a real bug exists 2) other underlying 
condition causes
                this to appear as a bug 3) The problem is well known to the 
GnuMachHurd pro-
                grammers and depends on conditions that will be repaired later 
4) something
                is wrong with my machine,installation or unknown changes that 
have happened
                during my installation (though I cannot think of any particular 
thing here).
                 5) Automatic procedures normally executed by a official-ready 
OS still 
                lacks in this version of Hurd,I do not know them,but I would 
have to set 
                them myself at this stage of kernel/OS development,therefore 
everything 
                looks strange to me.In other words: "I better stay in Nubia 
(where the 
                newbies live),wait for the programmers and enjoy the completely 
finished
                and "ready-for-sale" OS".


Log files:      I have read through the different log-files in /var/log/,but 
cannot find any
                particular message that explain what is wrong or that something 
is wrong.The 
                messages are simple and noninformatory as to this concern.


##############################################################################################

TRIES ON THE COMMAND LS -R IN ROOT DIRECTORY


Login:          ql sigbj    su sigbj    ql root   and   su root
Command:        cd /
Command:        ls -R |less
Result:         hos./dev/shm: Permission denied
                ls floppy : no Permission
                :/hurd/ext2fs:dummy.fs: No such file or directory
                ls : ./home/dummy :Translator died
                #But at the same time the directories are displayed
                #In addition there is noice in my machine telling that my 
harddisk is working
Comment:        I am not sure,but could it be some symbolic link pointing 
towards itself
                like '/usr ->.' that is causing this? I have experienced 
problems in SuSE linux
                and reported this to SuSE,where I found the screen was 
diplaying a continous
                rolling with the same directories coming back again and again 
fast.It was only
                the ls -R command causing this in SuSE and it is only the ls -R 
causing  in
                Hurd.In SuSE there was a directory in the distribution 7.0 that 
was symbolic
                linked to itself in /sbin/init/,I think.Now I can find no 
endless repeating
                rolling screen like in SuSE with Hurd.There is with 'ls -R' 
just a few seconds
                delay after a screen display,and then the whole recursive 
directory tree is
                displayed.  

                

##############################################################################################

This is all I have for whoever is interested in this way of testing out a 
system.

###############################################################################################