Re: Aw: Re: Generating 'cat' pages on make install

[Benoit Sigoure]

On Nov 5, 2007, at 11:08 AM, Jason Curl wrote:

> In the end, everything works like a dream. Here's my solution for  
> those who are interested, or who'd like to do something similar for  
> their projects.
>
> ----- configure.ac -----
> ...
> AC_PATH_PROG([RMAN], [rman])
> test x"$RMAN" = x && $ac_unset RMAN
> AM_MISSING_PROG([RMAN], [rman])
> ...
> ----- Makefile.am -----
> man_MANS = file1.3 file2.3
> html_DATA = ($man_MANS:.3=.html)
> BUILT_SOURCES = $(html_DATA)
> EXTRA_DIST = $(man_MANS) $(html_DATA)

BUILT_SOURCES are automatically distributed, so you can only list  
man_MANS in EXTRA_DIST.

> all-local: $(html_DATA)
> .3.html:
>     cd `dirname $^`; \
>     $(RMAN) --filter HTML -S `basename $^` > $@
> ----- end -----
>
> The 'cd' is required in case a man page is a reference to another  
> with the ".so" command, so "rman" find the page in case of VPATH  
> builds. It has the side effect of putting the results in the SOURCE  
> directory. Some people may not like this, but I believe it should  
> also work in the case of RO filesystems as the .html files are only  
> built when they're not available. As the directives above cause the  
> html files to be packaged, there should be no excuse, except for  
> filesystems with screwed up timestamps.

Yeah well, it's not the only case where you need a writable source  
tree.  For instance, texinfo sources already produce info files in  
the source tree.  The thing is, since they're distributed, if you're  
not a developer (so you're in the position "I want to deploy the  
software"), which is typically the the situation where the source  
tree might be read-only, it won't bug you since the tarball you use  
already contains the pre-built HTML pages.

Cheers,

--
Benoit Sigoure aka Tsuna
EPITA Research and Development Laboratory

PGP.sig
Description: This is a digitally signed message part