SVN bad behaviour makes what is written in the book wrong.

Daniel Shahaf danielsh at elego.de
Fri Mar 29 11:33:53 CDT 2013


C. Michael Pilato wrote on Fri, Mar 29, 2013 at 08:10:24 -0400:
> On 2/27/2013 2:35 PM, Daniel Shahaf wrote:
> > P.S.  *bump* to that issue about adding <link rel='canonical'> or
> > whatever the syntax was, so that people stop stumbling upon the
> > 8-years-old-and-counting book.
> 
> I hear you on the <link ...> request.  It's a decent idea (even though
> it's really /not /the appropriate use of the tag...), but I've yet to
> find a practical way to manage it automatically, especially when content
> has moved around between book versions (landing on differently named
> HTML pages as a result).

As a first-order approximation, old versions of the book could use the
one-giant-HTML-page version of the 1.7/1.8 book as their <link
rel='canonical'>.

I'm not sure what's the extent of the moving-around of content is.
Would "Manually set a custom XML tag on every <section2/> in every old
version of the book" be a feasible and semiautomable (i.e.,
automable for those <section2/>s that did _not_ move around, and manual
for the rest) process? [1]  The build process would then be patched to
render those tags as <link rel='canonical'/>.

Here, <section2> is a stand-in for whatever unit of book that becomes
a single HTML page in the live old HTML version of the book.

Daniel

[1] That's similar to how I did the svn_hash_gets search-and-replace
last week: I did a massive search-and-replace and then went through all
instances that were over 80 columns and fixed them.  Half of them could
be fixed with a single keystroke (one-off macros I bound to <F4> or
<F5>) and the others I did by hand.




More information about the svnbook-dev mailing list