Re: [patch 4/3] readline-6.2: Substitute inc-hist.texinfo

[Jan Kratochvil <jan dot kratochvil at redhat dot com>]

On Mon, 02 May 2011 16:34:51 +0200, Eli Zaretskii wrote:
> > --- a/readline/doc/hsuser.texi
> > +++ b/readline/doc/hsuser.texi
> > @@ -26,9 +26,10 @@ into another language, under the above conditions for modified versions.
> >  @node Using History Interactively
> >  @chapter Using History Interactively
> >  
> > -@ifclear BashFeatures
> > -@defcodeindex bt
> > -@end ifclear
> > +@c GDB bundling modification:
> > +@c @ifclear BashFeatures
> > +@c @defcodeindex bt
> > +@c @end ifclear
> 
> Why do we need to comment this out?

BTW it was already commented out the same way in readline-5.1.


> If BashFeatures is not set, the
> @defcodeindex line will not be in effect anyway.  Is something setting
> BashFeatures anywhere when we build our docs?

You are right BashFeatures is not set but this is @ifclear.  Therefore:
./../../readline/doc/hsuser.texi:31: Index `bt' already exists.


> >  This chapter describes how to use the @sc{gnu} History Library interactively,
> >  from a user's standpoint.  It should be considered a user's guide.  For
> >  information on using the @sc{gnu} History Library in your own programs,
> > -@pxref{Programming with GNU History}.
> > +@c GDB bundling modification:
> > +@pxref{Programming with GNU History, , , history, GNU History Library}.
> >  @end ifclear
> 
> Is this because the referenced node is only in the separate
> history.info manual, which is not part of hsuser.texi?

Yes:
./../../readline/doc/hsuser.texi:44: Cross reference to nonexistent node `Programming with GNU History' (perhaps incorrect sectioning?).


> If so, this is
> actually a bug in Readline docs which should be fixed upstream in a
> way that allows any project to include hsuser.texi without making such
> changes.  For example:
> 
>  @ifset HsUser
>  @pxref{Programming with GNU History, , , history, GNU History Library}.
>  @end ifset
>  @ifclear HsUser
>  @pxref{Programming with GNU History}.
>  @end ifclear
> 
> Then projects that include just hsuser.texi will need to pass the
> "-DHsUser" switch to makeinfo, while Readline will not.
> 
> I'm okay with making these changes for now, to allow people building
> GDB without failures.  But if Chet accepts the above change, we should
> switch to it in GDB, so that our doc/Makefile.in is ready for the next
> upstream release of Readline.

I guess defined BashFeatures was partially meant for some inclusion, OK.


Thanks,
Jan