jhalfs feedback

Jeremy Huntwork jhuntwork at linuxfromscratch.org
Wed Nov 16 08:22:14 PST 2005


Bruce Dubbs wrote:
> Jeremy Huntwork wrote:
> 
>>So what I'm proposing is that we add a switch for resuming a build
>>already started which will bypass a 'make clean' command. Otherwise the
>>default -M switch will first run 'make clean' on the build (removing
>>anything like user lfs, group lfs, /tools symlink, etc.) and then start
>>the build fresh.
> 
> 
> The system does not now automatically run any clean target.  The -M
> switch only runs `make` and that runs the target all which does not
> include clean*.

Yes, I think you misunderstood me. My proposal is to add that 
functionality. So that when you use jhalfs -M first 'make clean' is run, 
then 'make' unless you specify --resume or some such switch.


> When I first ran make, the system failed because of
> 020-creatingtoolsdir.  If the mkdir in that target were preceeded with a
> - or had a -p switch, it would not fail.  If the ln in the same target
> also had a leading - or had the switch -n added, it also would not fail.

Yes. What really should be done is not hard-code those chapter04 
commands in jhalfs, but have them sourced and run from the scriptlets 
like the rest of the book. In fact, as much as possible, jhalfs 
shouldn't hard-code anything. Granted there will likely always be 
work-arounds and exceptions to make the thing work, but I think we can 
find another solution here besides just dropping in those first few 
targets statically.

--
JH



More information about the alfs-discuss mailing list