[RFC] Using tidy to clean up profiles.

Kevin P. Fleming kpfleming at linuxfromscratch.org
Tue Jun 22 18:20:50 PDT 2004


Thomas Pegg wrote:

> As suggested by Kevin in a previous thread, I've looked into using tidy
> on the xml of the profiles to clean them up and standardize the
> formatting. I have commited a few examples in the BLFS profile. In
> paticular in chapter04 heimdal.xml and mitkrb.xml are two files, that
> have been run through tidy. I did it only a few files, before a change
> like this was committed profile-wide.

Those look very nice to me, that's what I was hoping tidy would be able 
to do.

> The entity reference issue is still a slight problem, the easiest way
> I've found to deal with it is sed the &'s back to an &. Around 95%
> of the time this works well except were there are actual &'s, in
> which case those have to be changed back manually, but since it's in a
> minority of the files, it's not a big problem. And really all files have
> to be run through tidy one time I think.

The solution I proposed before would work, but not until after the BLFS 
profile is converted to use XInclude. For profiles that are using 
XInclude, you can define an entity "real-amp" in general.ent that 
produces an ampersand (just like & does), and then use that instead. 
Tidy will mess with the leading ampersand of this entity reference too 
(since it's not a standard entity), but doing a sed from & back to & 
would fix it right back up, without any hand editing needed.

I don't know if this is really better, especially given that the BLFS 
profile hasn't been converted to XInclude yet and that's a pretty big 
job by itself. If someone was to run tidy, sed & back & and not 
realize there had originally been an & in that file, then xmllint 
will not validate the file and they will know something is wrong. That 
should be good enough.



More information about the alfs-discuss mailing list