Re: Documentation generated by doxygen and Debian Policy
Le Thu, Apr 19, 2012 at 05:02:13PM +0200, Thibaut Paumard a écrit :
> Le 19/04/12 16:20, Charles Plessy a écrit :
> >
> > I think that there is a consensus that generated files must be regenerable, and
> > that it is a bug if their build system is broken. The best way to ensure this
> > is to regenerate them at build time, however it is not a requirement. In
> > particular, I do not see advantages in distributing rebuilds that would for
> > instance only update time stamps. It is increasing debdiffs and heating the
> > planet for no benefit.
>
> File timestamps won't show up in debdiff. At the very least, the
> maintainer should (IMNSHO) regenerate the files himself and check that
> they match what ends-up being distributed. And he should also document
> this process in debian/README.source. Honestly, I think it's simpler and
> less error-prone to just distribute the regenerated files. But as you
> said, it's not a policy requirement.
There are other timestamps. Some documents contain the date where they were
built, as an indication for the last time they were modified. For the packages
I maintain, where I wrote manpages in DocBook XML, I do not regenerate them at
each build and store the generated nroff code in the VCS where the package is
maintained. This has the additional advantage that the package does not need
to build-depend on the DocBook toolchain. debian/rules contains a target to
rebuild, plus a comment explaining which packages to install.
The point I am trying to make is that there is no single rule that has to be
strictly followed by eveybody in all cases. Otherwise we would fall in the
same selective blindess as for the freeness of pictures, videos, music and
scientific data.
Cheers,
--
Charles Plessy
Debian Med packaging team,
http://www.debian.org/devel/debian-med
Tsurumi, Kanagawa, Japan
Reply to: