[blfs-dev] Two difficulties (one in all versions)

Bruce Dubbs bruce.dubbs at gmail.com
Wed Aug 21 15:18:04 PDT 2013

Fernando de Oliveira wrote:
> Did not know if sending here, to lfs -dev or to blfs -support.
> 1. Happened with lfs-7.2,7.3 (both svn) and now with 7.4-rc1.
> Long ago, I reported that each time I tried to chroot in LFS7.2 (LFS7.1
> as host), after logout and umount, had to reboot the system, because I
> could not get a terminal anymore (have a screen capture showing this,
> now), and today, X died, in the host.
> This has been reproduced, today, host LFS7.1, chroot in LFS7.4-rc1. I
> believe, now I understood the reason. I have (in 7.1):
> $ grep tty /etc/group
> tty:x:4:
> $ grep /dev/pts /etc/fstab
> devpts         /dev/pts     devpts   gid=4,mode=620      0     0
> But I was mounting and chrooting with gid 5, as per the book's instructions.
> Now, I have changed both to gid 5, to conform with the new versions, and
> the problem disappeared. Of course, had to change tape's gid to 4.
> Please, *warn me*, if, changing these, I am going to have any trouble in
> my old system.

I'm glad you figured it out.  We don't change this type of thing lightly 
and there was a fair amount of discussion back when it was done.

> 2. Happened, perhaps with all versions.
> Every time I try to create compressdoc with cat command, I obtain a
> corrupted unusable script. Then, use vim, instead, and copy/paste.
> Today, after the problem, just downloaded it from [1], after I saw, for
> the first time, I confess, the suggestion in blfs to do this, alternatively.
> Have produced a diff, in case anyone thinks this info is relevant.

I don't use compressdoc.  I have:

$ du -sh /opt/OpenJDK- /opt/kde-4.10.3/share/man \
          /opt/xorg  /share/man /opt/qt-4.8.4/share/man /usr/share/man
1.8M    /opt/OpenJDK-
228K    /opt/kde-4.10.3/share/man
17M     /opt/xorg/share/man
16K     /opt/qt-4.8.4/share/man
63M     /usr/share/man

That's about 2.1M on a 10G partition or about 0.02% How much space are 
you saving?

That said, there may be an issue with the size of the copy buffer.  When 
I try to select everything, I get a file of 10523 bytes, but when I do 
it in screen size blocks, the file is not corrupted and is 16870 bytes. 
  I would do an md5sum on the result, but a user's result may be 
different if a blank line is added or missing from multiple copy/paste 

Note too that when we are doing a copy from a browser, that it needs to 
remove html, so it may be a browser issue.

   -- Bruce

More information about the blfs-dev mailing list