Re: target hooks / plugins

[Joern Rennecke <amylaar@spamcop.net>]

Quoting "Joseph S. Myers" <joseph@codesourcery.com>:

> Please note that your initial change to implement automatic doc extraction
> should not result in any changes to the Texinfo content of the manual.
> Such fixes should all go in either before or after the automatic doc
> extraction change, but not at the same time; the doc extraction change
> should result in identical text in the manual, but with the Texinfo files
> produced in a different way.  I recommend sending such fixes before the
> automatic doc extraction change, since they do not depend on the FSF doing
> anything.

Duplicating all these changes separately by hand seems nigh impossible.
I think the best approach is then to take the auto-generated tm.texi as
the new tm.texi, and packages it up as a patch together with the
struct member / hook name changes that I made for consistency.

There is only one issue with using the current auto-generated tm.texi:
Unless special formatting was in force (e.g. @smallexample), I've removed
intra-paragraph newlines.  This should work in principle just as will
as with these newlines for producing output, but it looks somewhat daft
in tm.texi when you consider it as a source file.
Putting newlines in the input file would make it harder to read & edit,
since the documentation comes as C strings - and the GNU multiline string
extension has been deprecated some time ago.
I could pipe the documentation through fold -s, but that would also fold
extra-long lines outside the hook documentation, e.g.:
@@ -28,7 +28,8 @@
  @menu
  * Target Structure::    The @code{targetm} variable.
  * Driver::              Controlling how the driver runs the  
compilation passes.
-* Run-time Target::     Defining @samp{-m} options like  
@option{-m68000} and @option{-m68020}.
+* Run-time Target::     Defining @samp{-m} options like @option{-m68000} and
+@option{-m68020}.
  * Per-Function Data::   Defining data structures for per-function  
information.
  * Storage Layout::      Defining sizes and alignments of data.
  * Type Layout::         Defining sizes and properties of basic user  
data types.

So I suppose I'll have to add some fold mechanism into the documentation
output code.