[Makefile] PR bootstrap/12744 Allow release tarballs to build without bison/flex/makeinfo

Joseph S. Myers jsm@polyomino.org.uk
Tue Nov 4 22:04:00 GMT 2003


On Tue, 4 Nov 2003, Kelley Cook wrote:

> +ada/gnat_ug_unx.texi: ada/doctools/xgnatug$(exeext) \
> +		      $(srcdir)/ada/gnat_ug.texi $(srcdir)/ada/ug_words
> +	ada/doctools/xgnatug unx $(srcdir)/ada/gnat_ug.texi \
> +	  $(srcdir)/ada/ug_words \
> +	  ada/gnat_ug_unx.texi
> +	-$(INSRCDIR)cp -p $@ $(srcdir)/ada
[...]

> -$(srcdir)/f/intdoc.texi: $(INTDOC_DEPS) $(srcdir)/f/intdoc.in
> +f/intdoc.texi: $(INTDOC_DEPS) $(srcdir)/f/intdoc.in
>  	$(MAKE) f/intdoc$(build_exeext)
> -	f/intdoc$(build_exeext) > $(srcdir)/f/intdoc.texi
> +	f/intdoc$(build_exeext) > f/intdoc.texi
> +	-$(INSRCDIR)cp -p $@ $(srcdir)/f

The generated .texi files are in CVS so that update_web_docs works
(otherwise it would need to configure and build a fair amount of GCC).  
As they need to be regenerated and checked in whenever the source files
change, requiring maintainers to do a --enable-generated-files-in-srcdir
build whenever they change those files would seem error prone and liable
to make the versions in CVS more likely to get out of sync with the source
files.  A normal build from CVS (with timestamps made correct by
gcc_update) should never modify these files in the source directory.

-- 
Joseph S. Myers
jsm@polyomino.org.uk



More information about the Gcc-patches mailing list