Re: [opencbm-user] Errors compiling opencbm on Linux.

[Spiro T. <cb...@tr...>]

Hello Luca,

* On Sat, Nov 03, 2007 at 12:42:17AM +0100 gr...@gm... wrote:
 
> >Could you please check out the latest version on the
> >v0_4_0_bugfix_branch (tag: v0_4_0_bugfix_branch_make_showvars), enter
> >"makef -f LINUX/Makefile showvars-file", and send me the resulting
> >showvars.txt file? This might help me find out what is going wrong on
> >your system.
> 
> Ok, the file is here attached.

Thank you. Obviously, NOTHING works as I expected it.

Do you use the bash as your default shell, or another one?

 
> There must be some difference instead, since after having applied that 
> change above now the output of the "make -f LINUX/Makefile" command is:

Indeed, and that makes me very curious. I can only imagine that you are
using another default shell.

> >>  install: install-files
> >>-       test -z `type -p mkinfodir` || mkinfodir || /bin/true
> >>+       #test -z `type -p mkinfodir` || mkinfodir || /bin/true
> >
> >What was the point here? Why did you have to remove this line?
> 
> I have not the mkinfodir executable, nor i found a package it does 
> contain it, and it is not important (i guessed so at least :))

But I now belive this indicates exactly the problem you are having: The
line is written in a way that it should NOT execute mkinfodir if it is
not available! The "test -z `type -p mkinfodir'" line would be true if
mkinfodir does not exist. Nevertheless, on your machine, you get:

> For mkinfodir the error is:
> make[1]: Entering directory `/home/luca/opencbm/cbm4win/docs'
> gzip -c9 opencbm.info > /usr/local/info/opencbm.info.gz
> test -z `type -p mkinfodir` || mkinfodir || /bin/true
> test: 1: not: unexpected operator

Here, your machine seems to have a problem either with the backtick, or
with the type command. Now, type is a built-in of the bash, and the bash
can handle the back-tick, too! Thus, the mkinfodir command would never
have been executed.

Well, I looked at my /usr directory with "ls -ld /usr/*sh": Among
others, I get the following output:

$ ls -ld /bin/*sh
-rwxr-xr-x 1 root root 677184 2006-12-11 22:20 /bin/bash*
lrwxrwxrwx 1 root root      4 2007-08-02 21:43 /bin/sh -> bash*

Thus, here, sh is just a synonym for bash! So, now, I believe that on
your machine, sh does not execute bash, but the "real" sh shell. I
expect this to be the source of our problem.

So, could you please:

A. send me the output of the "ls -ld /usr/*sh" command on your machine?

B. if /usr/sh is not a symbolic link to /usr/bash, please rename the
   /bin/sh file, generate a symbolic link to /usr/bash, and retry to
   compile opencbm? Do you still have these problems?
   Don't forget to change the "original" sh back!

This might give me a clue if my assumption is right, and I would know in
which way I have to fix the problem.


> >>  Anyway, i issued this command;
> >>
> >>luca@c65:~/opencbm/cbm4win$ sudo insmod sys/linux/cbm.ko lp-0
> >>insmod: error inserting 'sys/linux/cbm.ko': -1 Unknown symbol in module
> >
> >"lp-0"? Why do you want to use this? Where do you have this option from?
> 
> At page 9 of the opencbm.pdf file, there is a "equal" sign so tight that 
> it seemed to me as a "dash", but looking to it closely it is an "equal" 
> symbol indeed, sorry for my mistake, but the font misleaded me :/

;)

Perhaps, this is the best proof why it is always the best to
copy-n-paste the command lines in problem or bug reports.

 
> Now that i given it the right parameter, and IT DOES WORK! GREAT!

Congratulation!

 
> The only problem is that i have to issue any command as 'root' user, but 
> that's no harm for now.

Do you use udev or not? For udev, "make install" should have copied the
file sys/linux/45-opencbm-parallel-rule into the directory
$(UDEV_RULES), that is, to /etc/udev/rules.d/. Is the file there?

If you are not using udev, than a simple "chmod 666 /dev/cbm" will fix
the problem by allowing EVERY user to use the device. You might want to
adjust the permissions in a more sensible way.


> Let me know if i could provide you any valuable 
> info on how to make the compilation working out of the box without the 
> hackish modifications i had to do.

Yes, I highly hope we can fix these issues.

Regards,
   Spiro.

-- 
Spiro R. Trikaliotis                              http://opencbm.sf.net/
http://www.trikaliotis.net/                     http://www.viceteam.org/