jhalfs feedback

Jeremy Huntwork jhuntwork at linuxfromscratch.org
Mon Nov 14 09:44:35 PST 2005

With Bruce's permission, forwarding the below message to alfs-discuss 
for review.



Bruce Dubbs wrote:
> Jeremy,
>   I'm not subscribed to alfs-discuss (perhaps I should be) but I want to
> give you some feedback.  These are just some random initial thoughts as
> I go through the system.  I have not yet run the generated Makefile.
> 1.  In functions, what is the variable BRW mean?  BROWN?  BOLD?  On my
> system it comes up bold.  The "RED" looks closer to brown than red on my
> system.  Perhaps you should have:
> BOLD= "ESC[0;1m"
> RED= "ESC[1;31m"    <-- Note the 1;
> instead of
> BRW= "^[[0;1m"
> RED= "^[[0;31m"

Sorry, that's me using silly variables. I first used this in the LiveCD
Makefile system, and when I was coding that using a black background,
that combination looked like bright white. :) Of course later I realized
that on a white background it shows as bold black. Obviously BOLD is a
better variable name. :)

> 2.  It took me a while to figure out what the script is doing.  Perhaps
> a paragraph or two in a README that gives a bit more detail is in order.
>  Particularly, what directories are used and what are the general
> contents/purpose of each?
> JHALFSDIR/{,commands,logs,lfs-development}
> BUILDDIR/{,sources}

Yes, we need better documentation all around. We can look this over
again before we release 0.2.

> 3.  The Makefile will fail early if the lfs user or group has already
> been created.  Some users (like me) would not like the clean-chapter4
> target removing the lfs user.  I note that it does not remove the lfs group.

I had noticed this too but have been too busy (or lazy!) to fix it yet.
Also, I think that 'userdel lfs' will get rid of the lfs group as well,
so long as no one else has been added to that group.

> 4.  The clean-chapter4 target will remove the lost+found directory if
> /mnt/lfs is a separate partition as recommended by the book.  Perhaps
> this could be fixed by just doing a mkdir to recreate it as it should be
> empty after the make clean, but there would need to be a test to see if
> it is a mounted partition first.

Ah, never noticed that. Thanks for the heads up.

> 5.  Perhaps its an LFS issue, but shouldn't all mkdir commands have the
> -p switch?

Well, certainly we can do that for any directories that jhalfs directly
makes. But as you implied, anything that's done via the commands in the
LFS book should be adjusted there first.

> Overall, the jhalfs system looks quite nice.  I especially like the
> ability to "discover" what it does fairly easily.  I can't say the same
> thing about a C program.  I also like the fact that does not require a
> "profile".
> Good job.

Thank you. :) It's getting to be a community piece of work now, too,
which is what it needs to be to improve. Thanks for the feedback.


More information about the alfs-discuss mailing list