> From: Paul Rogers <[email protected]>
> Date: Mon, 30 May 2016 14:11:31 -0700
> Subject: Re: [blfs-support] Dependency chain
>
> > I do not think this is reasonable for a couple of reasons.  First, we
> > do not have the manpower to spend the time to figure that out. Second,
> > it creates a maintenance nightmare because at every package update
> > (often 3-5 per day, sometimes more), we would potentially need to to
> > update the dependent packages for changes.
>
> Really?  OK.  I don't know your process.  I figured it'd be trivial--as
> you're updating each package, on each page you put a link to its
> dependencies, just "click on it" and add the anchor pointing back.
>


Core script is reasonably direct and clean - e.g.:
==
* parse the xml;
* get table of package/dep-categ/dep-label;
* use (repeated) sort/join cmds on orig+generated tables.
* extracted wanted data from relevant table(s)+row(s).
==


However, as is well known, there are - e.g.:
--
- circular deps;
- some 'blurring' of 'required'/'recommended'/'optional' dep-categories;
- some dep- sub-categories;
- deps within book and some outwith;
- blfs deps lists just omit entirely any ref to anything in lfs;
- and other 'real-world' messiness.
--


So in practice the above seq of 'naively'-generated tables, doesn't
hit a natural finite clean end-point; and one needs to augment the
'raw' data with some of one's own 'logic'/rules that you essentially
'patch-in' to the orig tbl.


(IOW, the real central issue is that the raw dep-logic is, as noted
often(-ish), not a simple tree or even a DAG.)



rgds,
akh





--
-- 
http://lists.linuxfromscratch.org/listinfo/blfs-support
FAQ: http://www.linuxfromscratch.org/blfs/faq.html
Unsubscribe: See the above information page

Reply via email to