RFC: Profiling and merging for the print edition

[Warren Block]

On Fri, 21 Jun 2013, Hiroki Sato wrote:

> Warren Block <wblock at wonkity.com> wrote
>  in <alpine.BSF.2.00.1306171145490.79915 at wonkity.com>:
>
> wb> My proposal for merging the ISBN branch and getting out of our
> wb> two-branch problem.
> wb>
> wb> First, profiling works.  A demo file of the Porter's Handbook is here:
> wb> http://wonkity.com/~wblock/profiledemo/
> wb>
> wb> After applying the patch, build either the online or print version of
> wb> the book with:
> wb>
> wb>   make EDITION=online clean book.html
> wb>   make EDITION=print  clean book.html
> wb>
> wb> The "online" version includes all sections with edition="online" or an
> wb> edition not set.
>
> Please do not add a separate make variable for that.  We already have
> FORMATS, and setting profiling variables (profile.attribute and
> profile.value in XSLT) based on it should work.

You mean a new format should be added, like "pdf-withoutonline"?  That 
sounds like a better option.

> Also, we should consider/discuss what kind of difference we need
> actually and should allow between the two first.  Basically I am
> against marking up everything with "online" or "print" because it
> just makes the source files complicated.  I think it is useful only
> for separating lengthy preface, edition history, and pgpkey list from
> the body in book.xml.

Agreed.  The experiment showed that only sections which should not be 
included in the printed book could be marked with edition="online". 
All other sections would be left with edition unset.

There is a complication.  Links to sections which have been eliminated 
by profiling will break, so those would also need profiling.  I don't 
know how big an issue that will be, but suspect it will not be very 
much.

Hopefully, the profiling will be unobtrusive enough so that it can be 
done in head.  That will make turnaround time for the next printed 
version of the Handbook much shorter.