Beta website

Anderson Lizardo lizardo at
Sun Feb 13 12:30:17 PST 2005

On Sunday 13 February 2005 14:28, Matthew Burgess wrote:
> now has the stuff that myself and
> Jeremy Huntwork have been working on.  It's basically serving a working
> copy of the 'www2' repository available on belgarath.

> IIUC this will be 
> changed slightly, so the working copy is actually stored elsewhere, and
> then just the necessary bits 'rsync'ed over to the
> directory.

Hm, I remember seeing this somewhere... :P

> Aside from the obvious lack of content, there are some things listed in
> that need looking at.  In
> particular, the render-* scripts will need to render into the
> www2 working copy, then perform an 'svn commit' to save the changes into
> the website repo.  This way, if something does come along and wipe the
> website out, we can just re-checkout the www2 repository and it's all
> there immediately - no need to remember which branches of which books
> need to be regenerated.

I personally don't like the idea of commiting to the website repos the result 
of every book rendering/hint/patch. Let me explain why:

- It takes a _long_ time to download the repos (I'm still checking it out, to 
have an idea).
- We are just duplicating things. Hints/Patches just need to be "svn update" 
from their repos to recreate missing parts. Books can be re-rendered.
- I know there are some "static" parts on the website (those which we had to 
recover when the website was deleted), like the stable renderings/tarballs, 
BLFS artwork. I think a site-wide beckup system (like the one Jeremy Utley 
was working on, not sure if it's currently running) is more efficient then 
storing on SVN.

There are some other but I think you got the idea.

> There's a 'scripts' directory in the www2 repo.  I'd like it if all the
> render* scripts and the post-commit script can be put in there, so we
> can keep track of changes to them.  Obviously this needn't be served out
> by Apache, so the rsync command will have to have it added to its ignore
> list.

Nice idea. Can be even implemented on our current site.

Regarding the website layout/structure:

- I like it, it's clean, but has some alignment bugs on Konqueror.

- I noticed you used SSI to merge things. If you plan to go with this, we will 
have to add this requirement to mirrors. PHP includes can be used here, and 
we can do a "post-processing" (basically call php on the command line) so the 
content that goes to /home/httpd/ has only HTML and 
mirrors don't need to have PHP installed.

- I think at least the SVN changelogs should stay.

BTW, just finished downloading the repos:

37M     lfs-website2

It's a little fat ;-)
Anderson Lizardo
lizardo at

More information about the website mailing list