Flow Chart

Joachim Beckers jbeckers at linuxfromscratch.org
Sun Feb 6 08:13:20 PST 2005


Alexander Lang wrote:
> I don't know how it is done in nALFS since i never really looked at the
> code,

I wasn't really talking about the code, but see below...

> but i was kinda following what Boris (NeoCool) was writing:
> he had plugins for different "syntaxes", such as the LFS-Book or the
> nALFS syntax. What these plugins did was parsing the profile-files into
> a profile-structure. This structure consisted of a name a version and a 
> double linked list of chapters. Each Chapter had a name and a double 
> linked list of packages. The package structure had all the information,
> such as downloads, build commands, dependencies,.....

Sounds exactly like what I had in mind and to some extent like nALFS 
too, but more important: it's the exact structure of a profile too. See 

> I don't think it is harder to edit such a structure than the 
> profile-files,  and it gives us more flexibility.

Basically, it all depends on th ui you give the user. However, there's 
two big reasons why we should go for editing the profile-files:
1) The ui-program doesn't know about the internal datastructure of 
alfsd. It does nothing but controlling the build, sending over profiles 
(as XML, see 2) and monitoring what happens on the other side of the 
2) The protocol between alfsd and the ui will be XML-based (at least 
that's what most of us would like to see). Hence there will be no way to 
send over some kind of datastructure form the ui to the daemon.


More information about the alfs-discuss mailing list