jhalfs and nALFS

Jeremy Huntwork jhuntwork at lightcubesolutions.com
Mon Apr 16 08:02:35 PDT 2012

On 4/16/12 10:47 AM, jon wrote:
> would some mind telling the true story of the /downfall/ of nALFS?
> Clearly a lot of love time and attention went into it and now its just
> lying there.
> Personally jhalfs feels completely wrong. I love when it works and there
> it is a fully built system but it so happens that last time I could
> only confidently do this with LFS-6.3 which left me doing all sorts of
> weird things to keep this hybrid 'code-frankenstein's monster' on the
> road.

There were two main reason nALFS was abandoned and jhalfs was implemented.

First, nALFS required maintaining a separate set of XML configs to feed 
into it as build recipes. We wanted something that could extract 
commands directly from the book and use those to build, so there was 
less chance of human error and (hopefully) a greater chance of achieving 
a pristine 'by-the-book' build.

Secondly, at the time, there weren't a lot of experienced developers 
contributing to the project, so it was felt that writing "simpler" 
scripts in shell to do all the heavy lifting would encourage more people 
to help contribute and maintain. This appears to have worked, for a time.

Unfortunately, over time jhalfs has also become a little heavy and 
unwieldy. It would be nice to take some of jhalfs' core concepts and 
refactor them into a more elegant solution.


More information about the lfs-chat mailing list