Re: dtd version

[Patrice Dumas]

On Tue, Dec 18, 2012 at 01:30:47AM +0000, Karl Berry wrote:
> Sorry, I didn't notice this before:
> 
> # version for the dtd.  In between releases, PACKAGE_VERSION is not set
> # to a value that corresponds to actual files on the internet.
> # TEXINFO_DTD_VERSION=$PACKAGE_VERSION
> TEXINFO_DTD_VERSION=5.00
> AC_SUBST([TEXINFO_DTD_VERSION])
> 
> Well, 5.00 doesn't correspond to anything on the Internet either.
> But anyway, it was always my intention to put pretest dtd's on the
> net.  I did it in the past for many versions, just failed to do so for
> these, but I can from now on.

The comment is a bit misleading in fact.  The reasons for not using the 
PACKAGE_VERSION directly are that there is no need to change the DTD 
version for each (pre-)release since the DTD may change at a different 
pace, and also that each change of the DTD means a need to regenerate
test case results, so having it change with each pretest would be quite
inconvenient.

In any case, the fact that it is a DTD not already present on the internet
is a good thing, it means that the user will have to use the one that
comes from the sources (CVS version or tarball) which is good in case 
the DTD changes between releases.  So the comment is misleading as it 
should instead be along

# version for the dtd.  Use the next DTD version to avoid using a 
# DTD from internet that wouldn't be in sync, and also to avoid 
# unnnecessary changes in XML output file headers.

So I think that I would prefer if you did _not_ put the dtd file on 
the internet for the pretest...


> Overall, it makes more sense to me to use $PACKAGE_VERSION, because the
> DTD for 4.13.9x isn't necessarily the same as what we'll end up with at
> 5.0, after all ... is there some other reason not to?

In most cases, the DTD should be in a frozen state when entering pretest
phases.  So it should be more likely not to have changes in DTD.  But
changes may still happen, because of a bugfix in the DTD or a last
minute change as I did for incode and mergeindex.

Still, my point of view is that, as long as the new dtd file is only in 
the source and not on the internet, it doesn't matter that the dtd 
changes between a pretest and the release, because the one in the 
sources will always be in sync.

-- 
Pat